Introduction to Pandas.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1899,
   "metadata": {},
   "outputs": [],
   "source": [
    "!git add \"Introduction\\ to\\ Pandas.ipynb\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1900,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[master 9daa228] updating Pandas introduction\n",
      " 1 file changed, 490 insertions(+), 18 deletions(-)\n"
     ]
    }
   ],
   "source": [
    "!git commit -m \"updating Pandas introduction\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1901,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Enumerating objects: 5, done.\n",
      "Counting objects: 100% (5/5), done.\n",
      "Delta compression using up to 8 threads\n",
      "Compressing objects: 100% (3/3), done.\n",
      "Writing objects: 100% (3/3), 2.28 KiB | 777.00 KiB/s, done.\n",
      "Total 3 (delta 2), reused 0 (delta 0)\n",
      "remote: Resolving deltas: 100% (2/2), completed with 2 local objects.\u001b[K\n",
      "To https://github.com/Luc-Bertin/TDs_ESILV.git\n",
      "   7f17ae7..9daa228  master -> master\n"
     ]
    }
   ],
   "source": [
    "!git push origin master"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1254,
   "metadata": {},
   "outputs": [],
   "source": [
    "from IPython.core.interactiveshell import InteractiveShell\n",
    "InteractiveShell.ast_node_interactivity = \"all\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Discover Pandas \n",
    "\n",
    "Pandas is built on top of Numpy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First, check the overview of the package [here](https://pandas.pydata.org/docs/getting_started/overview.html)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    ">`pandas.DataFrame` is a 2-dimensional labeled data structure with columns of potentially different types. You can think of it like a spreadsheet or SQL table.\n",
    "It got rows and columns' labels and can greatly contain and handle missing data.\n",
    "\n",
    "Dealing with initially less structured, clean and complete data consists in most of the time spent by the data scientist."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'0.25.3'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "pd?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "14.7 ns ± 0.386 ns per loop (mean ± std. dev. of 7 runs, 100000000 loops each)\n"
     ]
    }
   ],
   "source": [
    "%%timeit\n",
    "3+2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the rest of this tutorial we will mainly work on the `DataFrame` class, although we first have to introduce 2 other core data structures provided by the package: the `Series` and the `Index`, as they are each constitutive of `DataFrame` and the former share very similar API with the DataFrame class."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Series"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "one-dimensional array of indexed data. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1007,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    3\n",
       "1    2\n",
       "2    1\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1007,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.Series([3,2,1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "with explicit index definition !"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Example:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1008,
   "metadata": {},
   "outputs": [],
   "source": [
    "serie_1 = pd.Series([3,2,1], index=[93,129, 219394])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1009,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "93        3\n",
       "129       2\n",
       "219394    1\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1009,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie_1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1010,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Int64Index([93, 129, 219394], dtype='int64')"
      ]
     },
     "execution_count": 1010,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie_1.index"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "a dictionnary-like, object with possible keys repetition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1011,
   "metadata": {},
   "outputs": [],
   "source": [
    "serie = pd.Series([3,2,1], index=[\"rené\", \"rené\", \"jean\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1012,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "rené    3\n",
       "rené    2\n",
       "jean    1\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1012,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1013,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([3, 2, 1])"
      ]
     },
     "execution_count": 1013,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1014,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['rené', 'rené', 'jean'], dtype='object')"
      ]
     },
     "execution_count": 1014,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie.index"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Access by key"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1015,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "rené    3\n",
       "rené    2\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1015,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie['rené']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Set a new key pair"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1016,
   "metadata": {},
   "outputs": [],
   "source": [
    "serie['joseph'] = 5"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Change a value for a key"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1017,
   "metadata": {},
   "outputs": [],
   "source": [
    "serie['rené'] = 4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1018,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "rené      4\n",
       "rené      4\n",
       "jean      1\n",
       "joseph    5\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1018,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1019,
   "metadata": {},
   "outputs": [],
   "source": [
    "serie['rené'] = [4,3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1020,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "rené      4\n",
       "rené      3\n",
       "jean      1\n",
       "joseph    5\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1020,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* delete a key val pair"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1021,
   "metadata": {},
   "outputs": [],
   "source": [
    "del serie[\"rené\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1022,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "jean      1\n",
       "joseph    5\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1022,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1023,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "jean    1\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1023,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie[0:4:2] # indexing: not possible in a simple dict "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* lookup "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1024,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "False\n",
      "True\n"
     ]
    }
   ],
   "source": [
    "print('rené' in serie)\n",
    "print(\"jean\" in serie)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- When index is unique, pandas use a hashtable just like `dict`s : O(1). \n",
    "- When index is non-unique and sorted, pandas use binary search O(logN)\n",
    "- When index is non-unique and not-sorted pandas need to check all the keys just like a list look-up: O(N).\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "using a `dict` in the `pd.Series` constructor automatically assigns the index as the ordered keys in the `dict` (for Python 3.6 and later though, the index is in the same order as the insertion order).\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1025,
   "metadata": {},
   "outputs": [],
   "source": [
    "test = pd.Series(dict(zip([\"ea\",\"fzf\",\"aeif\"], [2,3,2])))\n",
    "# with zip or using a dict\n",
    "test2 = pd.Series({\"ea\":2, \"fzf\":3, \"aeif\":2}, index=[\"ea\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1026,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "aeif    2\n",
       "ea      2\n",
       "fzf     3\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1026,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1027,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ea    2\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1027,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**If multiple different types reside in a Series, all of the data will get upcasted to a dtype that accommodates all of the data involved.**\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1030,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ea    2\n",
       "dtype: object"
      ]
     },
     "execution_count": 1030,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test2 = pd.Series({\"ea\":2, \"fzf\":3, \"aeif\":\"zf\"}, index=[\"ea\"])\n",
    "test2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1033,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ea    2.0\n",
       "dtype: float64"
      ]
     },
     "execution_count": 1033,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test2 = pd.Series({\"ea\":2, \"fzf\":3, \"aeif\":2.4}, index=[\"ea\"])\n",
    "test2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**dtype=object** means that the best common type infered representation for the contents of the `pd.Series` is that they are Python objects."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "this also mean performance decreases, **any operations on the data will be done at the Python level** (1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## selection in Series"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1038,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "aeif    False\n",
       "ea      False\n",
       "fzf      True\n",
       "dtype: bool"
      ]
     },
     "execution_count": 1038,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(test>2) # return a boolean array we can later use to filter the values we want (returning True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1039,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "aeif    True\n",
       "ea      True\n",
       "fzf     True\n",
       "dtype: bool"
      ]
     },
     "execution_count": 1039,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(test<4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1041,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "aeif    False\n",
       "ea      False\n",
       "fzf      True\n",
       "dtype: bool"
      ]
     },
     "execution_count": 1041,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# not \"and\" but \"&\" : & operator is a bitwise \"and\"\n",
    "(test>2) & (test < 4) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1042,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "pandas.core.series.Series"
      ]
     },
     "execution_count": 1042,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "type((test>2) & (test < 4) )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1043,
   "metadata": {},
   "outputs": [],
   "source": [
    "# mask ( the last expression whose result is an pd.Serie stored in the variable mask)\n",
    "mask = (test>2) & (test < 4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1044,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "fzf    3\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1044,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test[mask]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1046,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ea     2\n",
       "fzf    3\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1046,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# fancy indexing (<=> selecting multiple indexes using a list of indexes)\n",
    "test[[\"ea\", \"fzf\"]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1048,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "aeif    2\n",
       "ea      2\n",
       "fzf     3\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1048,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# explicit index slicing (using the labels of the indexes)\n",
    "test[\"aeif\": \"fzf\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1049,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "aeif    2\n",
       "ea      2\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1049,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# implicit index slicing (using integers)\n",
    "test[0: 2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- using explicit indexes while slicing ***include*** the final index"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- using implicit index in slicing ***exclude*** the final index"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "What about i defined **explicit integer indexes at first** and i want to slice ? 🙄"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## using loc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1061,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1     4\n",
       "2     8\n",
       "3    51\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1061,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie2 = pd.Series({1:4, 2:8, 3:51})\n",
    "serie2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1076,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "51"
      ]
     },
     "execution_count": 1076,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "3    51\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1076,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie2[3] # indexing: defaults to select explicit index /  with label 3\n",
    "serie2[2:3] # slicing: defaults to select implicit index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1085,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4"
      ]
     },
     "execution_count": 1085,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "2     8\n",
       "3    51\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1085,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie2.loc[1] # indexing: on explicit index\n",
    "serie2.loc[2:3] # slicing: on explicit index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1086,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8"
      ]
     },
     "execution_count": 1086,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "3    51\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1086,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie2.iloc[1] # indexing: on implicit index\n",
    "serie2.iloc[2:3] # slicing: on implicit index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1090,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1     4\n",
       "2     8\n",
       "3    51\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1090,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "2     8\n",
       "3    51\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1090,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie2.loc[1:5] # slicing: explicit index \n",
    "serie2.iloc[1:5] # slicing: implicit index "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1094,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    4\n",
       "2    8\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1094,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "2     8\n",
       "3    51\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1094,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie2.loc[[1,2]] # fancy indexing\n",
    "serie2.iloc[[1,2]] # fancy indexing "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Index object "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* are immutable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1096,
   "metadata": {},
   "outputs": [
    {
     "ename": "TypeError",
     "evalue": "Index does not support mutable operations",
     "output_type": "error",
     "traceback": [
      "Traceback \u001b[0;36m(most recent call last)\u001b[0m:\n",
      "  File \u001b[1;32m\"<ipython-input-1096-707f9cda8675>\"\u001b[0m, line \u001b[1;32m1\u001b[0m, in \u001b[1;35m<module>\u001b[0m\n    serie2.index[0]=18\n",
      "\u001b[0;36m  File \u001b[0;32m\"/Users/lucbertin/.pyenv/versions/3.5.7/lib/python3.5/site-packages/pandas/core/indexes/base.py\"\u001b[0;36m, line \u001b[0;32m4260\u001b[0;36m, in \u001b[0;35m__setitem__\u001b[0;36m\u001b[0m\n\u001b[0;31m    raise TypeError(\"Index does not support mutable operations\")\u001b[0m\n",
      "\u001b[0;31mTypeError\u001b[0m\u001b[0;31m:\u001b[0m Index does not support mutable operations\n"
     ]
    }
   ],
   "source": [
    "serie2.index[0]=18"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* can be sliced or indexed (just like an array)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1104,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 1104,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie2.index[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1105,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Int64Index([1, 2], dtype='int64')"
      ]
     },
     "execution_count": 1105,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie2.index[:2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1106,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Int64Index([1], dtype='int64')"
      ]
     },
     "execution_count": 1106,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie2.index & {1, 5}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1107,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Int64Index([2, 3, 5], dtype='int64')"
      ]
     },
     "execution_count": 1107,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie2.index ^ {1,5}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# DataFrame"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* sequence of \"aligned\" Series objects (sharing same indexes / like an Excel file )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* each Series object is a column"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Hence `pd.DataFrame` can be seen as dictionnary of Series objects"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Flexible rows and columns' labels (`Index` objects for both)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1205,
   "metadata": {},
   "outputs": [],
   "source": [
    "serie1 = pd.Series({\"Luc\": 25, \"Corentin\":29, \"René\": 40})\n",
    "serie2 = pd.Series({\"René\": \"100%\", \"Corentin\": \"25%\", \"Luc\": \"20%\"})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1206,
   "metadata": {},
   "outputs": [],
   "source": [
    "# dictionnary of pd.Series\n",
    "df = pd.DataFrame({\"note\": serie1, \n",
    "                   \"charge_de_travail\": serie2})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1207,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>charge_de_travail</th>\n",
       "      <th>note</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Corentin</th>\n",
       "      <td>25%</td>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Luc</th>\n",
       "      <td>20%</td>\n",
       "      <td>25</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>René</th>\n",
       "      <td>100%</td>\n",
       "      <td>40</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         charge_de_travail  note\n",
       "Corentin               25%    29\n",
       "Luc                    20%    25\n",
       "René                  100%    40"
      ]
     },
     "execution_count": 1207,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1208,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['Corentin', 'Luc', 'René'], dtype='object')"
      ]
     },
     "execution_count": 1208,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "Index(['charge_de_travail', 'note'], dtype='object')"
      ]
     },
     "execution_count": 1208,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# index objects on both columns and rows\n",
    "df.index\n",
    "df.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> If you pass an index and / or columns, you are guaranteeing the index and / or columns of the resulting DataFrame. Thus, a dict of Series plus a specific index will discard all data not matching up to the passed index."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1209,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>note</th>\n",
       "      <th>autre</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Corentin</th>\n",
       "      <td>29.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Luc</th>\n",
       "      <td>25.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Julie</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          note autre\n",
       "Corentin  29.0   NaN\n",
       "Luc       25.0   NaN\n",
       "Julie      NaN   NaN"
      ]
     },
     "execution_count": 1209,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2 = pd.DataFrame({\"note\": serie1, \n",
    "                    \"charge_de_travail\": serie2}, \n",
    "                   index=[\"Corentin\", \"Luc\", \"Julie\"],\n",
    "                   columns=[\"note\", \"autre\"])\n",
    "df2 \n",
    "# filled with NaN (\"Not A Number\") \n",
    "# when no value exist for the given (row_index, column_index)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1210,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3, 2)"
      ]
     },
     "execution_count": 1210,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "shape: tuple of the number of elements with respect to each dimension"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For a 1D array, the shape would be (n,) where n is the number of elements in your array.\n",
    "\n",
    "For a 2D array, the shape would be (n,m) where n is the number of rows and m is the number of columns in your array"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "accessing a column/`Serie` by key : "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1211,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Corentin    29\n",
       "Luc         25\n",
       "René        40\n",
       "Name: note, dtype: int64"
      ]
     },
     "execution_count": 1211,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['note']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "Using the attribute notation is not advised for assignements as some methods or attributes of the same name already exist in the DataFrame class' own namespace"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1212,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Corentin    29\n",
       "Luc         25\n",
       "René        40\n",
       "Name: note, dtype: int64"
      ]
     },
     "execution_count": 1212,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.note"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The `DataFrame` can be constructed using a list of dictionary\n",
    "each dict element is a row\n",
    "each key of each dict refers a column"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1213,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>c</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.0</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>3</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     a  b    c\n",
       "0  1.0  2  NaN\n",
       "1  NaN  3  4.0"
      ]
     },
     "execution_count": 1213,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2 = pd.DataFrame([{'a': 1, 'b': 2}, {'b': 3, 'c': 4}])\n",
    "df2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1214,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>c</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Jean</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Jacques</th>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>René</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         a  b  c\n",
       "Jean     1  1  3\n",
       "Jacques  1  2  4\n",
       "René     1  1  1"
      ]
     },
     "execution_count": 1214,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame([(1, 1, 3), (1, 2,4), (1,1,1)],\n",
    "             columns=[\"a\", \"b\", \"c\"],\n",
    "            index=[\"Jean\", \"Jacques\", \"René\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Indexing works the same way as for Series, but you have to account this time for the second dimension"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`df.loc_or_iloc[ dim1 = rows, dim2 = columns]`\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1215,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>charge_de_travail</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Corentin</th>\n",
       "      <td>25%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Luc</th>\n",
       "      <td>20%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>René</th>\n",
       "      <td>100%</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         charge_de_travail\n",
       "Corentin               25%\n",
       "Luc                    20%\n",
       "René                  100%"
      ]
     },
     "execution_count": 1215,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.iloc[:3, :1] # implicit indexing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**columns slicing/indexing** is optional here, without specifying it, you **select only rows** "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1216,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>charge_de_travail</th>\n",
       "      <th>note</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Corentin</th>\n",
       "      <td>25%</td>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Luc</th>\n",
       "      <td>20%</td>\n",
       "      <td>25</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>René</th>\n",
       "      <td>100%</td>\n",
       "      <td>40</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         charge_de_travail  note\n",
       "Corentin               25%    29\n",
       "Luc                    20%    25\n",
       "René                  100%    40"
      ]
     },
     "execution_count": 1216,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.iloc[:3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1217,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>charge_de_travail</th>\n",
       "      <th>note</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Corentin</th>\n",
       "      <td>25%</td>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Luc</th>\n",
       "      <td>20%</td>\n",
       "      <td>25</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         charge_de_travail  note\n",
       "Corentin               25%    29\n",
       "Luc                    20%    25"
      ]
     },
     "execution_count": 1217,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.loc[\"Corentin\":\"Luc\",\"charge_de_travail\":\"note\"] # explicit indexing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "same thing here, **only rows selected**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1218,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>charge_de_travail</th>\n",
       "      <th>note</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Corentin</th>\n",
       "      <td>25%</td>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         charge_de_travail  note\n",
       "Corentin               25%    29"
      ]
     },
     "execution_count": 1218,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.loc[:\"Corentin\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1219,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>charge_de_travail</th>\n",
       "      <th>note</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Corentin</th>\n",
       "      <td>25%</td>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Luc</th>\n",
       "      <td>20%</td>\n",
       "      <td>25</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         charge_de_travail  note\n",
       "Corentin               25%    29\n",
       "Luc                    20%    25"
      ]
     },
     "execution_count": 1219,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.loc[[\"Corentin\", \"Luc\"], :] # mixing slicing and fancy indexing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Something to mention here, by default, without using indexers like `loc` and `iloc`:\n",
    "- indexing directly `df`, performs the indexing on its columns **(ex:1)**\n",
    "- slicing by conditions, or using a slice notation (::), is performed on rows **(ex:2)**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1220,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>charge_de_travail</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Corentin</th>\n",
       "      <td>25%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Luc</th>\n",
       "      <td>20%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>René</th>\n",
       "      <td>100%</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         charge_de_travail\n",
       "Corentin               25%\n",
       "Luc                    20%\n",
       "René                  100%"
      ]
     },
     "execution_count": 1220,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[[\"charge_de_travail\"]] # indexing directly df defaults to columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "(2) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1221,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Corentin     True\n",
       "Luc         False\n",
       "René        False\n",
       "Name: charge_de_travail, dtype: bool"
      ]
     },
     "execution_count": 1221,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mask = df[\"charge_de_travail\"]==\"25%\" \n",
    "mask"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1222,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>charge_de_travail</th>\n",
       "      <th>note</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Corentin</th>\n",
       "      <td>25%</td>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         charge_de_travail  note\n",
       "Corentin               25%    29"
      ]
     },
     "execution_count": 1222,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[mask] # masking directly df is operated on rows"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1223,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>charge_de_travail</th>\n",
       "      <th>note</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Corentin</th>\n",
       "      <td>25%</td>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Luc</th>\n",
       "      <td>20%</td>\n",
       "      <td>25</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>René</th>\n",
       "      <td>100%</td>\n",
       "      <td>40</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         charge_de_travail  note\n",
       "Corentin               25%    29\n",
       "Luc                    20%    25\n",
       "René                  100%    40"
      ]
     },
     "execution_count": 1223,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[:3] # slicing directly df is operated on rows"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Operations on Pandas"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Element-wise operations are made easy in `Pandas`.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* 3 - 2 <=> substract(3,2) <=> binary operation (2 inputs)\n",
    "* -2 <=> neg(2) <=> unary operation (one input)\n",
    "* sin(2) <=> unary operation (one input)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "in `Pandas` : \n",
    "  - unary operations on `df`s elements preserve the indexes.\n",
    "  - binary operations on elements from 2 `df`s align the operations on the indexes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1224,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1225,
   "metadata": {},
   "outputs": [],
   "source": [
    "rng = np.random.RandomState(42) # for reproducibility\n",
    "data = rng.randint(0,10, (3,4)) # creating an array of random integer values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1226,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6</td>\n",
       "      <td>9</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>7</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   0  1  2  3\n",
       "0  6  3  7  4\n",
       "1  6  9  2  6\n",
       "2  7  4  3  7"
      ]
     },
     "execution_count": 1226,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.DataFrame(data)\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1227,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>9</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "      <td>9</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   0  1  2  3\n",
       "0  7  2  5  4\n",
       "1  1  7  5  1\n",
       "2  4  0  9  5\n",
       "3  8  0  9  2"
      ]
     },
     "execution_count": 1227,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2 = pd.DataFrame(rng.randint(0,10, (4,4)))\n",
    "df2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1228,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>9</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "      <td>9</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   0  1  2  3\n",
       "1  1  7  5  1\n",
       "0  7  2  5  4\n",
       "2  4  0  9  5\n",
       "3  8  0  9  2"
      ]
     },
     "execution_count": 1228,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2 = df2.reindex([1,0,2,3]) #just to show rearranged indexes (does not change the association with the indexed data)\n",
    "df2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1229,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>13.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>8.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>11.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>12.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      0     1     2     3\n",
       "0  13.0   5.0  12.0   8.0\n",
       "1   7.0  16.0   7.0   7.0\n",
       "2  11.0   4.0  12.0  12.0\n",
       "3   NaN   NaN   NaN   NaN"
      ]
     },
     "execution_count": 1229,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df + df2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "on line of index 0, `7+6 = 13`\n",
    " which shows indexes had been aligned during the binary operation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "also notice the union of the indices during the binary operation. If one may not exist in either of the dataframes and the result can't be evalutated, `NaN` fill the concerned entries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1230,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>13.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>8.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>11.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>12.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>33.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>34.0</td>\n",
       "      <td>27.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      0     1     2     3\n",
       "0  13.0   5.0  12.0   8.0\n",
       "1   7.0  16.0   7.0   7.0\n",
       "2  11.0   4.0  12.0  12.0\n",
       "3  33.0  25.0  34.0  27.0"
      ]
     },
     "execution_count": 1230,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.__add__(df2, fill_value=25) # used in the binary operation 25+8 = 33)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Operation between pandas series and a pandas dataframe"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the Numpy Docs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Broadcasting is **how numpy treats arrays with different shapes during arithmetic operations**. \n",
    "Subject to certain constraints, the smaller array is “broadcast” across the larger array so that they **have compatible shapes.**\n",
    "Broadcasting provides a means of vectorizing array operations so that looping occurs in C instead of Python"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The only requirement for broadcasting is a way aligning array dimensions such that either :\n",
    "* aligned dimensions are equal (so that operations are done on an element-by-element basis from 2 arrays of same shape)\n",
    "* one of the aligned dimensions is 1 (in other words, dimensions with size 1 are stretched or “copied” to match the dimension of the other array)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Operations between `pandas.Series` and `pandas.DataFram` respect the numpy broadcasting rules:\n",
    ">  If the two arrays **differ in their number of dimensions**, the shape of the one with **fewer dimensions is padded with ones on its leading (left) side.'**  (2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1231,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((3, 4), (4,), (1, 4))"
      ]
     },
     "execution_count": 1231,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape, df.iloc[1].shape, df.iloc[1][np.newaxis, :].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1232,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6</td>\n",
       "      <td>9</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>7</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   0  1  2  3\n",
       "0  6  3  7  4\n",
       "1  6  9  2  6\n",
       "2  7  4  3  7"
      ]
     },
     "execution_count": 1232,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "0    6\n",
       "1    9\n",
       "2    2\n",
       "3    6\n",
       "Name: 1, dtype: int64"
      ]
     },
     "execution_count": 1232,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df\n",
    "df.iloc[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1233,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>-6</td>\n",
       "      <td>5</td>\n",
       "      <td>-2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>-5</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   0  1  2  3\n",
       "0  0 -6  5 -2\n",
       "1  0  0  0  0\n",
       "2  1 -5  1  1"
      ]
     },
     "execution_count": 1233,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df - df.iloc[1] #row-wise (1,4) copied other 3 times => (3,4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1234,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>-6</td>\n",
       "      <td>5</td>\n",
       "      <td>-2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>-5</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   0  1  2  3\n",
       "0  0 -6  5 -2\n",
       "1  0  0  0  0\n",
       "2  1 -5  1  1"
      ]
     },
     "execution_count": 1234,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df - df.iloc[1].sample(4) # again: kept the index alignements during computation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "if you want to do it columnwise and not row wise"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1236,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>-3.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>-2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-7.0</td>\n",
       "      <td>-3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     0    1    2    3\n",
       "0  0.0 -3.0  1.0 -2.0\n",
       "1 -3.0  0.0 -7.0 -3.0\n",
       "2  5.0  2.0  1.0  5.0\n",
       "3  NaN  NaN  NaN  NaN"
      ]
     },
     "execution_count": 1236,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.__sub__(df.iloc[1], axis=0) # caution, the indexes operations will be based on the column indexes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1237,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>0</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6</td>\n",
       "      <td>9</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>7</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   a  b  0  d\n",
       "0  6  3  7  4\n",
       "1  6  9  2  6\n",
       "2  7  4  3  7"
      ]
     },
     "execution_count": 1237,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.columns = [\"a\",\"b\",0,\"d\"]\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1238,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "a    6\n",
       "b    9\n",
       "0    2\n",
       "d    6\n",
       "Name: 1, dtype: int64"
      ]
     },
     "execution_count": 1238,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.iloc[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1240,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>0</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>a</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>d</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     a    b    0    d\n",
       "0  4.0  1.0  5.0  2.0\n",
       "1  NaN  NaN  NaN  NaN\n",
       "2  NaN  NaN  NaN  NaN\n",
       "a  NaN  NaN  NaN  NaN\n",
       "b  NaN  NaN  NaN  NaN\n",
       "d  NaN  NaN  NaN  NaN"
      ]
     },
     "execution_count": 1240,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.__sub__(df.iloc[1], axis=0) \n",
    "# based on the column indexes\n",
    "# only 0 match with one of the column index label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1247,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((3,), (3, 4))"
      ]
     },
     "execution_count": 1247,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[0].shape, df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1248,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-2</td>\n",
       "      <td>-15</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    0  a   b   d\n",
       "0 NaN  5   1 NaN\n",
       "1 NaN  2   4 NaN\n",
       "2 NaN -2 -15 NaN"
      ]
     },
     "execution_count": 1248,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2 = df - pd.DataFrame([(1,2), (4,5), (9,19)], columns=[\"a\",\"b\"])\n",
    "df2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "### dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1903,
   "metadata": {},
   "outputs": [],
   "source": [
    "s = pd.Series(['The', 3, 'brown', 'fox'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1904,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0        TheThe\n",
       "1             6\n",
       "2    brownbrown\n",
       "3        foxfox\n",
       "dtype: object"
      ]
     },
     "execution_count": 1904,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s*2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1905,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['The', 3, 'brown', 'fox'], dtype=object)"
      ]
     },
     "execution_count": 1905,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s.values"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> [Link](https://stackoverflow.com/questions/29877508/what-does-dtype-object-mean-while-creating-a-numpy-array). Creating an array with dtype=object is different. The memory taken by the array now is filled with pointers to Python objects which are being stored elsewhere in memory (much like a Python list is really just a list of pointers to objects, not the objects themselves)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1902,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "a         int64\n",
      "b         int64\n",
      "0         int64\n",
      "d         int64\n",
      "notes    object\n",
      "dtype: object\n",
      "Name          object\n",
      "id_account    object\n",
      "id_client     object\n",
      "dtype: object\n"
     ]
    }
   ],
   "source": [
    "print(df.dtypes)\n",
    "print(df2.dtypes) \n",
    "# NaN is a floating-point value, \n",
    "# hence the Series embedding it gets its dtype upcasted to float (if it were an int)\n",
    "# this pd.Series supports fast operations contrarily to a Series of dtype=object\n",
    "# because Python needs to type check dynamically every time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 863,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.14 ms ± 142 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n"
     ]
    }
   ],
   "source": [
    "%timeit np.arange(1E6, dtype=\"int\").sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 864,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "77.8 ms ± 4.15 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n"
     ]
    }
   ],
   "source": [
    "%timeit np.arange(1E6, dtype=\"object\").sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 865,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.25 ms ± 17.7 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n"
     ]
    }
   ],
   "source": [
    "%timeit np.arange(1E6, dtype=\"float\").sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Managing missing values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1250,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    False\n",
       "1     True\n",
       "dtype: bool"
      ]
     },
     "execution_count": 1250,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.Series([2, np.nan]).isnull()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1251,
   "metadata": {},
   "outputs": [],
   "source": [
    "df2.iloc[0,2] = np.nan"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1252,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-2</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    0  a     b   d\n",
       "0 NaN  5   NaN NaN\n",
       "1 NaN  2   4.0 NaN\n",
       "2 NaN -2 -15.0 NaN"
      ]
     },
     "execution_count": 1252,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      0      a      b     d\n",
       "0  True  False   True  True\n",
       "1  True  False  False  True\n",
       "2  True  False  False  True"
      ]
     },
     "execution_count": 1252,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2\n",
    "df2.isnull()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1255,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    2.0\n",
       "dtype: float64"
      ]
     },
     "execution_count": 1255,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.Series([2, np.nan]).dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1257,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-2</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    0  a     b   d\n",
       "0 NaN  5   NaN NaN\n",
       "1 NaN  2   4.0 NaN\n",
       "2 NaN -2 -15.0 NaN"
      ]
     },
     "execution_count": 1257,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>a</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   a\n",
       "0  5\n",
       "1  2\n",
       "2 -2"
      ]
     },
     "execution_count": 1257,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [0, a, b, d]\n",
       "Index: []"
      ]
     },
     "execution_count": 1257,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-2</td>\n",
       "      <td>-15.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   a     b\n",
       "0  5   NaN\n",
       "1  2   4.0\n",
       "2 -2 -15.0"
      ]
     },
     "execution_count": 1257,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>a</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   a\n",
       "0  5\n",
       "1  2\n",
       "2 -2"
      ]
     },
     "execution_count": 1257,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2\n",
    "df2.dropna(axis=1) # drop a column when contains one NA value\n",
    "df2.dropna(axis=0) # drop a row when contains one NA value\n",
    "df2.dropna(axis=1, how=\"all\") # drop a column when contains all NA value\n",
    "df2.dropna(axis=1, thresh=3) # drop a column if below 3 non-NA value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1259,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-2</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    0  a     b   d\n",
       "0 NaN  5   NaN NaN\n",
       "1 NaN  2   4.0 NaN\n",
       "2 NaN -2 -15.0 NaN"
      ]
     },
     "execution_count": 1259,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.0</td>\n",
       "      <td>5</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2.0</td>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>-2</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     0  a     b    d\n",
       "0  2.0  5   2.0  2.0\n",
       "1  2.0  2   4.0  2.0\n",
       "2  2.0 -2 -15.0  2.0"
      ]
     },
     "execution_count": 1259,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-2</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    0  a     b   d\n",
       "0 NaN  5   4.0 NaN\n",
       "1 NaN  2   4.0 NaN\n",
       "2 NaN -2 -15.0 NaN"
      ]
     },
     "execution_count": 1259,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-2.0</td>\n",
       "      <td>-2.0</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     0    a     b   d\n",
       "0  5.0  5.0   NaN NaN\n",
       "1  2.0  2.0   4.0 NaN\n",
       "2 -2.0 -2.0 -15.0 NaN"
      ]
     },
     "execution_count": 1259,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2\n",
    "df2.fillna(value=2) #fill NA with specified value\n",
    "# fill NA backwards \n",
    "# i.e. using the following non-null element\n",
    "# to fill preceding NA ones\n",
    "# defaults on rows basis\n",
    "df2.fillna(method=\"bfill\") \n",
    "df2.fillna(method=\"bfill\", axis=1) # on column basis"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### MultiIndex"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1260,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "group1  Luc        18\n",
       "        René        4\n",
       "        Seb        17\n",
       "group2  Alex        4\n",
       "        Camille     2\n",
       "        Jean       23\n",
       "group3  Sophie     25\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1260,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = {('group1', 'Luc'): 18,\n",
    "        ('group2', 'Jean'): 23,\n",
    "        ('group1', 'Seb'): 17,\n",
    "        ('group1', 'René'): 4,\n",
    "        ('group2', 'Alex'): 4,\n",
    "        ('group3', 'Sophie'): 25,\n",
    "        ('group2', 'Camille'): 2 }\n",
    "serie = pd.Series(data)\n",
    "serie"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1261,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "group1    18\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1261,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie[:,\"Luc\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1271,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Luc     18\n",
       "René     4\n",
       "Seb     17\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1271,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie[\"group1\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1273,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "group1  Luc       18\n",
       "group2  Jean      23\n",
       "group3  Sophie    25\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1273,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie[serie>=18]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1274,
   "metadata": {},
   "outputs": [],
   "source": [
    "# creating the multi-index using cartesian product\n",
    "index = pd.MultiIndex.from_arrays([['group1', 'a', 'b', 'b'], [\"Luc\", 2, 1, 2]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1268,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "group1  Luc    18.0\n",
       "a       2       NaN\n",
       "b       1       NaN\n",
       "        2       NaN\n",
       "dtype: float64"
      ]
     },
     "execution_count": 1268,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie.reindex(index) # works for multi-index too !\n",
    "# Conform Series to new index with optional filling logic, placing\n",
    "# NA/NaN in locations having no value in the previous index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1269,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr:last-of-type th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>subject</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Bob</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Guido</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Sue</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>type</th>\n",
       "      <th>HR</th>\n",
       "      <th>Temp</th>\n",
       "      <th>HR</th>\n",
       "      <th>Temp</th>\n",
       "      <th>HR</th>\n",
       "      <th>Temp</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>year</th>\n",
       "      <th>visit</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">2013</th>\n",
       "      <th>1</th>\n",
       "      <td>52.0</td>\n",
       "      <td>36.4</td>\n",
       "      <td>38.0</td>\n",
       "      <td>36.6</td>\n",
       "      <td>32.0</td>\n",
       "      <td>38.1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>28.0</td>\n",
       "      <td>37.7</td>\n",
       "      <td>47.0</td>\n",
       "      <td>35.4</td>\n",
       "      <td>50.0</td>\n",
       "      <td>36.4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">2014</th>\n",
       "      <th>1</th>\n",
       "      <td>30.0</td>\n",
       "      <td>37.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>36.6</td>\n",
       "      <td>49.0</td>\n",
       "      <td>37.7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>52.0</td>\n",
       "      <td>36.8</td>\n",
       "      <td>31.0</td>\n",
       "      <td>35.5</td>\n",
       "      <td>36.0</td>\n",
       "      <td>37.6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "subject      Bob       Guido         Sue      \n",
       "type          HR  Temp    HR  Temp    HR  Temp\n",
       "year visit                                    \n",
       "2013 1      52.0  36.4  38.0  36.6  32.0  38.1\n",
       "     2      28.0  37.7  47.0  35.4  50.0  36.4\n",
       "2014 1      30.0  37.0  16.0  36.6  49.0  37.7\n",
       "     2      52.0  36.8  31.0  35.5  36.0  37.6"
      ]
     },
     "execution_count": 1269,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# hierarchical indices and columns\n",
    "index = pd.MultiIndex.from_product([[2013, 2014], [1, 2]],\n",
    "                                   names=['year', 'visit'])\n",
    "columns = pd.MultiIndex.from_product(\n",
    "    [['Bob', 'Guido', 'Sue'], ['HR', 'Temp']],names=['subject', 'type'])\n",
    "# mock some data\n",
    "data = np.round(np.random.randn(4, 6), 1)\n",
    "data[:, ::2] *= 10\n",
    "data += 37\n",
    "# create the DataFrame\n",
    "health_data = pd.DataFrame(data, index=index, columns=columns) \n",
    "health_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1280,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>type</th>\n",
       "      <th>HR</th>\n",
       "      <th>Temp</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>year</th>\n",
       "      <th>visit</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">2013</th>\n",
       "      <th>1</th>\n",
       "      <td>52.0</td>\n",
       "      <td>36.4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>28.0</td>\n",
       "      <td>37.7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "type          HR  Temp\n",
       "year visit            \n",
       "2013 1      52.0  36.4\n",
       "     2      28.0  37.7"
      ]
     },
     "execution_count": 1280,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "health_data.loc[:2013 , (\"Bob\")]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1282,
   "metadata": {},
   "outputs": [],
   "source": [
    "#health_data.loc[(:,1),[\"Bob\"]] # can't use the tuple to define index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1285,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr:last-of-type th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>subject</th>\n",
       "      <th>Bob</th>\n",
       "      <th>Guido</th>\n",
       "      <th>Sue</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>type</th>\n",
       "      <th>HR</th>\n",
       "      <th>HR</th>\n",
       "      <th>HR</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>year</th>\n",
       "      <th>visit</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2013</th>\n",
       "      <th>1</th>\n",
       "      <td>52.0</td>\n",
       "      <td>38.0</td>\n",
       "      <td>32.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2014</th>\n",
       "      <th>1</th>\n",
       "      <td>30.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>49.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "subject      Bob Guido   Sue\n",
       "type          HR    HR    HR\n",
       "year visit                  \n",
       "2013 1      52.0  38.0  32.0\n",
       "2014 1      30.0  16.0  49.0"
      ]
     },
     "execution_count": 1285,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "idx = pd.IndexSlice\n",
    "health_data.loc[idx[:, 1], idx[:, 'HR']]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Unstacking and Stacking: a matter of dimnesionality"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Creating a multiIndex rather than a simlpe Index is like **creating an extra-dimension in our dataset.**\n",
    "\n",
    "We can take for **each year**, a **2D sub-dataframe** composed of **Bob's HR visits**.\n",
    "\n",
    "This DataFrame hence can be seen as having **4 dimensions.**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "we can go back and forth from a multi-index series to a dataframe using unstack, so that one of the index level occupies the extra dimension given by the transition to a `DataFrame`\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1290,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "group1  Luc        18\n",
       "        René        4\n",
       "        Seb        17\n",
       "group2  Alex        4\n",
       "        Camille     2\n",
       "        Jean       23\n",
       "group3  Sophie     25\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1290,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1291,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Alex</th>\n",
       "      <th>Camille</th>\n",
       "      <th>Jean</th>\n",
       "      <th>Luc</th>\n",
       "      <th>René</th>\n",
       "      <th>Seb</th>\n",
       "      <th>Sophie</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>group1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>18.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>17.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>group2</th>\n",
       "      <td>4.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>23.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>group3</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>25.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        Alex  Camille  Jean   Luc  René   Seb  Sophie\n",
       "group1   NaN      NaN   NaN  18.0   4.0  17.0     NaN\n",
       "group2   4.0      2.0  23.0   NaN   NaN   NaN     NaN\n",
       "group3   NaN      NaN   NaN   NaN   NaN   NaN    25.0"
      ]
     },
     "execution_count": 1291,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie.unstack() #level -1 by default = most inner one"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1299,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>group1</th>\n",
       "      <th>group2</th>\n",
       "      <th>group3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Alex</th>\n",
       "      <td>NaN</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Camille</th>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Jean</th>\n",
       "      <td>NaN</td>\n",
       "      <td>23.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Luc</th>\n",
       "      <td>18.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>René</th>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Seb</th>\n",
       "      <td>17.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Sophie</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>25.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         group1  group2  group3\n",
       "Alex        NaN     4.0     NaN\n",
       "Camille     NaN     2.0     NaN\n",
       "Jean        NaN    23.0     NaN\n",
       "Luc        18.0     NaN     NaN\n",
       "René        4.0     NaN     NaN\n",
       "Seb        17.0     NaN     NaN\n",
       "Sophie      NaN     NaN    25.0"
      ]
     },
     "execution_count": 1299,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df3 = serie.unstack(level=0)\n",
    "df3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1300,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>index</th>\n",
       "      <th>group1</th>\n",
       "      <th>group2</th>\n",
       "      <th>group3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Alex</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Camille</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Jean</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Luc</td>\n",
       "      <td>18.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>René</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Seb</td>\n",
       "      <td>17.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Sophie</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>25.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     index  group1  group2  group3\n",
       "0     Alex     NaN     4.0     NaN\n",
       "1  Camille     NaN     2.0     NaN\n",
       "2     Jean     NaN    23.0     NaN\n",
       "3      Luc    18.0     NaN     NaN\n",
       "4     René     4.0     NaN     NaN\n",
       "5      Seb    17.0     NaN     NaN\n",
       "6   Sophie     NaN     NaN    25.0"
      ]
     },
     "execution_count": 1300,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# to reset the index and create it as a simple new column you can use reset_index()\n",
    "df3.reset_index()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can do some aggregation by index level (we are going to see this extensively on `GroupBy` section\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1307,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr:last-of-type th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th>subject</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Bob</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Guido</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Sue</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>type</th>\n",
       "      <th>HR</th>\n",
       "      <th>Temp</th>\n",
       "      <th>HR</th>\n",
       "      <th>Temp</th>\n",
       "      <th>HR</th>\n",
       "      <th>Temp</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>year</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2013</th>\n",
       "      <td>40.0</td>\n",
       "      <td>37.05</td>\n",
       "      <td>42.5</td>\n",
       "      <td>36.00</td>\n",
       "      <td>41.0</td>\n",
       "      <td>37.25</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2014</th>\n",
       "      <td>41.0</td>\n",
       "      <td>36.90</td>\n",
       "      <td>23.5</td>\n",
       "      <td>36.05</td>\n",
       "      <td>42.5</td>\n",
       "      <td>37.65</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "subject   Bob        Guido          Sue       \n",
       "type       HR   Temp    HR   Temp    HR   Temp\n",
       "year                                          \n",
       "2013     40.0  37.05  42.5  36.00  41.0  37.25\n",
       "2014     41.0  36.90  23.5  36.05  42.5  37.65"
      ]
     },
     "execution_count": 1307,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr:last-of-type th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th>subject</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Bob</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Guido</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Sue</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>type</th>\n",
       "      <th>HR</th>\n",
       "      <th>Temp</th>\n",
       "      <th>HR</th>\n",
       "      <th>Temp</th>\n",
       "      <th>HR</th>\n",
       "      <th>Temp</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>visit</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>41.0</td>\n",
       "      <td>36.70</td>\n",
       "      <td>27.0</td>\n",
       "      <td>36.60</td>\n",
       "      <td>40.5</td>\n",
       "      <td>37.9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>40.0</td>\n",
       "      <td>37.25</td>\n",
       "      <td>39.0</td>\n",
       "      <td>35.45</td>\n",
       "      <td>43.0</td>\n",
       "      <td>37.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "subject   Bob        Guido          Sue      \n",
       "type       HR   Temp    HR   Temp    HR  Temp\n",
       "visit                                        \n",
       "1        41.0  36.70  27.0  36.60  40.5  37.9\n",
       "2        40.0  37.25  39.0  35.45  43.0  37.0"
      ]
     },
     "execution_count": 1307,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>type</th>\n",
       "      <th>HR</th>\n",
       "      <th>Temp</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>year</th>\n",
       "      <th>visit</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">2013</th>\n",
       "      <th>1</th>\n",
       "      <td>40.666667</td>\n",
       "      <td>37.033333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>41.666667</td>\n",
       "      <td>36.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">2014</th>\n",
       "      <th>1</th>\n",
       "      <td>31.666667</td>\n",
       "      <td>37.100000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>39.666667</td>\n",
       "      <td>36.633333</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "type               HR       Temp\n",
       "year visit                      \n",
       "2013 1      40.666667  37.033333\n",
       "     2      41.666667  36.500000\n",
       "2014 1      31.666667  37.100000\n",
       "     2      39.666667  36.633333"
      ]
     },
     "execution_count": 1307,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "health_data.mean(level='year')\n",
    "health_data.mean(level='visit')\n",
    "health_data.mean(axis=1, level='type')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### pd.concat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1311,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.0</td>\n",
       "      <td>False</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2.0</td>\n",
       "      <td>False</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>False</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     0      a    b    d\n",
       "0  2.0  False  2.0  2.0\n",
       "1  2.0  False  0.0  2.0\n",
       "2  2.0  False  0.0  2.0"
      ]
     },
     "execution_count": 1311,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-2</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    0  a     b   d\n",
       "0 NaN  5   NaN NaN\n",
       "1 NaN  2   4.0 NaN\n",
       "2 NaN -2 -15.0 NaN"
      ]
     },
     "execution_count": 1311,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1\n",
    "df2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1318,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-2</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     0  a     b    d\n",
       "0  2.0  0   2.0  2.0\n",
       "1  2.0  0   0.0  2.0\n",
       "2  2.0  0   0.0  2.0\n",
       "0  NaN  5   NaN  NaN\n",
       "1  NaN  2   4.0  NaN\n",
       "2  NaN -2 -15.0  NaN"
      ]
     },
     "execution_count": 1318,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.0</td>\n",
       "      <td>False</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2.0</td>\n",
       "      <td>False</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>False</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-2</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     0      a    b    d   0  a     b   d\n",
       "0  2.0  False  2.0  2.0 NaN  5   NaN NaN\n",
       "1  2.0  False  0.0  2.0 NaN  2   4.0 NaN\n",
       "2  2.0  False  0.0  2.0 NaN -2 -15.0 NaN"
      ]
     },
     "execution_count": 1318,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.concat([df1, df2], axis=0) # concatenate rows (default)\n",
    "pd.concat([df1, df2], axis=1) # concatenate columns (default)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "the indices are preserved, even duplicated"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`verify_integrity=True` can check if index from each df are differents"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`ignore_index=True` just override the indexes after concatenation by a new integer one"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`keys = [\"source1\", \"source2\"]` leave the indexes as is but create a new outer level from the 2 different sources/df of the data concatenated"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1320,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"3\" valign=\"top\">source1</th>\n",
       "      <th>0</th>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"3\" valign=\"top\">source2</th>\n",
       "      <th>0</th>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-2</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             0  a     b    d\n",
       "source1 0  2.0  0   2.0  2.0\n",
       "        1  2.0  0   0.0  2.0\n",
       "        2  2.0  0   0.0  2.0\n",
       "source2 0  NaN  5   NaN  NaN\n",
       "        1  NaN  2   4.0  NaN\n",
       "        2  NaN -2 -15.0  NaN"
      ]
     },
     "execution_count": 1320,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.concat([df1, df2], axis=0, keys=[\"source1\", \"source2\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`join='inner'` keeps only the columns in common from the concatenation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1333,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "      <th>note</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-2</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    0  a     b   d  note\n",
       "0 NaN  5   NaN NaN     2\n",
       "1 NaN  2   4.0 NaN     2\n",
       "2 NaN -2 -15.0 NaN     2"
      ]
     },
     "execution_count": 1333,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2[\"note\"] = 2\n",
    "df2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1336,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-2</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     0  a     b    d\n",
       "0  2.0  0   2.0  2.0\n",
       "1  2.0  0   0.0  2.0\n",
       "2  2.0  0   0.0  2.0\n",
       "0  NaN  5   NaN  NaN\n",
       "1  NaN  2   4.0  NaN\n",
       "2  NaN -2 -15.0  NaN"
      ]
     },
     "execution_count": 1336,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.concat([df1, df2], axis=0, join='inner')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1342,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Corentin    29\n",
       "Luc         25\n",
       "René        40\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1342,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "Corentin     25%\n",
       "Luc          20%\n",
       "René        100%\n",
       "dtype: object"
      ]
     },
     "execution_count": 1342,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "Corentin      29\n",
       "Luc           25\n",
       "René          40\n",
       "Corentin     25%\n",
       "Luc          20%\n",
       "René        100%\n",
       "dtype: object"
      ]
     },
     "execution_count": 1342,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "serie1\n",
    "serie2\n",
    "serie1.append(serie2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1343,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/lucbertin/.pyenv/versions/3.5.7/lib/python3.5/site-packages/pandas/core/frame.py:7138: FutureWarning: Sorting because non-concatenation axis is not aligned. A future version\n",
      "of pandas will change to not sort by default.\n",
      "\n",
      "To accept the future behavior, pass 'sort=False'.\n",
      "\n",
      "To retain the current behavior and silence the warning, pass 'sort=True'.\n",
      "\n",
      "  sort=sort,\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>a</th>\n",
       "      <th>b</th>\n",
       "      <th>d</th>\n",
       "      <th>note</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>4.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-2</td>\n",
       "      <td>-15.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     0  a     b    d  note\n",
       "0  2.0  0   2.0  2.0   NaN\n",
       "1  2.0  0   0.0  2.0   NaN\n",
       "2  2.0  0   0.0  2.0   NaN\n",
       "0  NaN  5   NaN  NaN   2.0\n",
       "1  NaN  2   4.0  NaN   2.0\n",
       "2  NaN -2 -15.0  NaN   2.0"
      ]
     },
     "execution_count": 1343,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1.append(df2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Merge"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1371,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Amount</th>\n",
       "      <th>Name</th>\n",
       "      <th>accountNumber</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10000</td>\n",
       "      <td>LIVRET A</td>\n",
       "      <td>AC1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>109300</td>\n",
       "      <td>Compte Épargne Retraite</td>\n",
       "      <td>AC2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2984</td>\n",
       "      <td>Quadretto</td>\n",
       "      <td>AC3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1999</td>\n",
       "      <td>Compte Courant</td>\n",
       "      <td>AC4</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Amount                     Name accountNumber\n",
       "0   10000                 LIVRET A           AC1\n",
       "1  109300  Compte Épargne Retraite           AC2\n",
       "2    2984                Quadretto           AC3\n",
       "3    1999           Compte Courant           AC4"
      ]
     },
     "execution_count": 1371,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Name</th>\n",
       "      <th>id_account</th>\n",
       "      <th>id_client</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Luc</td>\n",
       "      <td>AC1</td>\n",
       "      <td>ID1099</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>René</td>\n",
       "      <td>AC2</td>\n",
       "      <td>ID1091</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Jean</td>\n",
       "      <td>AC3</td>\n",
       "      <td>ID1018</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Jean</td>\n",
       "      <td>AC4</td>\n",
       "      <td>ID1018</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Joseph</td>\n",
       "      <td>AC5</td>\n",
       "      <td>ID1021</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Name id_account id_client\n",
       "0     Luc        AC1    ID1099\n",
       "1    René        AC2    ID1091\n",
       "2    Jean        AC3    ID1018\n",
       "3    Jean        AC4    ID1018\n",
       "4  Joseph        AC5    ID1021"
      ]
     },
     "execution_count": 1371,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_account = pd.DataFrame({'accountNumber': [\"AC1\", \"AC2\", \"AC3\", \"AC4\"],\n",
    "                   'Amount': [10000, 109300, 2984, 1999],\n",
    "                   'Name': [\"LIVRET A\", \"Compte Épargne Retraite\", \"Quadretto\", \"Compte Courant\"]})\n",
    "df_client = pd.DataFrame({'id_account': [\"AC1\", \"AC2\", \"AC3\", \"AC4\", \"AC5\"],\n",
    "                   'Name': [\"Luc\", \"René\", \"Jean\", \"Jean\", \"Joseph\"],\n",
    "                   'id_client': [\"ID1099\", \"ID1091\", \"ID1018\", \"ID1018\", \"ID1021\"]})\n",
    "df_account\n",
    "df_client"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1372,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Amount</th>\n",
       "      <th>Name_x</th>\n",
       "      <th>accountNumber</th>\n",
       "      <th>Name_y</th>\n",
       "      <th>id_account</th>\n",
       "      <th>id_client</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10000</td>\n",
       "      <td>LIVRET A</td>\n",
       "      <td>AC1</td>\n",
       "      <td>Luc</td>\n",
       "      <td>AC1</td>\n",
       "      <td>ID1099</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>109300</td>\n",
       "      <td>Compte Épargne Retraite</td>\n",
       "      <td>AC2</td>\n",
       "      <td>René</td>\n",
       "      <td>AC2</td>\n",
       "      <td>ID1091</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2984</td>\n",
       "      <td>Quadretto</td>\n",
       "      <td>AC3</td>\n",
       "      <td>Jean</td>\n",
       "      <td>AC3</td>\n",
       "      <td>ID1018</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1999</td>\n",
       "      <td>Compte Courant</td>\n",
       "      <td>AC4</td>\n",
       "      <td>Jean</td>\n",
       "      <td>AC4</td>\n",
       "      <td>ID1018</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Amount                   Name_x accountNumber Name_y id_account id_client\n",
       "0   10000                 LIVRET A           AC1    Luc        AC1    ID1099\n",
       "1  109300  Compte Épargne Retraite           AC2   René        AC2    ID1091\n",
       "2    2984                Quadretto           AC3   Jean        AC3    ID1018\n",
       "3    1999           Compte Courant           AC4   Jean        AC4    ID1018"
      ]
     },
     "execution_count": 1372,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.merge(left=df_account, right=df_client, \n",
    "         left_on=\"accountNumber\", \n",
    "         right_on=\"id_account\",\n",
    "         how='inner')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1375,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Amount</th>\n",
       "      <th>Name_account</th>\n",
       "      <th>accountNumber</th>\n",
       "      <th>Name_client</th>\n",
       "      <th>id_account</th>\n",
       "      <th>id_client</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10000.0</td>\n",
       "      <td>LIVRET A</td>\n",
       "      <td>AC1</td>\n",
       "      <td>Luc</td>\n",
       "      <td>AC1</td>\n",
       "      <td>ID1099</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>109300.0</td>\n",
       "      <td>Compte Épargne Retraite</td>\n",
       "      <td>AC2</td>\n",
       "      <td>René</td>\n",
       "      <td>AC2</td>\n",
       "      <td>ID1091</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2984.0</td>\n",
       "      <td>Quadretto</td>\n",
       "      <td>AC3</td>\n",
       "      <td>Jean</td>\n",
       "      <td>AC3</td>\n",
       "      <td>ID1018</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1999.0</td>\n",
       "      <td>Compte Courant</td>\n",
       "      <td>AC4</td>\n",
       "      <td>Jean</td>\n",
       "      <td>AC4</td>\n",
       "      <td>ID1018</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Joseph</td>\n",
       "      <td>AC5</td>\n",
       "      <td>ID1021</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Amount             Name_account accountNumber Name_client id_account  \\\n",
       "0   10000.0                 LIVRET A           AC1         Luc        AC1   \n",
       "1  109300.0  Compte Épargne Retraite           AC2        René        AC2   \n",
       "2    2984.0                Quadretto           AC3        Jean        AC3   \n",
       "3    1999.0           Compte Courant           AC4        Jean        AC4   \n",
       "4       NaN                      NaN           NaN      Joseph        AC5   \n",
       "\n",
       "  id_client  \n",
       "0    ID1099  \n",
       "1    ID1091  \n",
       "2    ID1018  \n",
       "3    ID1018  \n",
       "4    ID1021  "
      ]
     },
     "execution_count": 1375,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_merged = pd.merge(left=df_account, right=df_client, \n",
    "         left_on=\"accountNumber\", \n",
    "         right_on=\"id_account\",\n",
    "         how='right',\n",
    "         suffixes=[\"_account\", \"_client\"])\n",
    "df_merged"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1376,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Amount</th>\n",
       "      <th>Name_account</th>\n",
       "      <th>accountNumber</th>\n",
       "      <th>Name_client</th>\n",
       "      <th>id_client</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10000.0</td>\n",
       "      <td>LIVRET A</td>\n",
       "      <td>AC1</td>\n",
       "      <td>Luc</td>\n",
       "      <td>ID1099</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>109300.0</td>\n",
       "      <td>Compte Épargne Retraite</td>\n",
       "      <td>AC2</td>\n",
       "      <td>René</td>\n",
       "      <td>ID1091</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2984.0</td>\n",
       "      <td>Quadretto</td>\n",
       "      <td>AC3</td>\n",
       "      <td>Jean</td>\n",
       "      <td>ID1018</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1999.0</td>\n",
       "      <td>Compte Courant</td>\n",
       "      <td>AC4</td>\n",
       "      <td>Jean</td>\n",
       "      <td>ID1018</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Joseph</td>\n",
       "      <td>ID1021</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Amount             Name_account accountNumber Name_client id_client\n",
       "0   10000.0                 LIVRET A           AC1         Luc    ID1099\n",
       "1  109300.0  Compte Épargne Retraite           AC2        René    ID1091\n",
       "2    2984.0                Quadretto           AC3        Jean    ID1018\n",
       "3    1999.0           Compte Courant           AC4        Jean    ID1018\n",
       "4       NaN                      NaN           NaN      Joseph    ID1021"
      ]
     },
     "execution_count": 1376,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# to drop the (same) column we have been merging on\n",
    "df_merged.drop('id_account', axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1482,
   "metadata": {},
   "outputs": [],
   "source": [
    "#df_notes[\"eleve\"] = (df_notes\n",
    "# .eleve\n",
    "# .astype(\"category\")\n",
    "# .cat.rename_categories(\n",
    "#     new_categories=\n",
    "#        [\"eleve{}\".format(i) for i in range(df_notes.eleve.nunique())]\n",
    "# )\n",
    "#)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Apply "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For the following sections, we are going to use real world data of students'grades from an exam I gave 😜 The data has been anonymised to fit GDPR regulation."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It has been retrieved by scrapping automatically the online web app that stores the results from each passed quizz."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You will see along the way we will need to make multiple modifications to our original data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1720,
   "metadata": {},
   "outputs": [],
   "source": [
    "# !curl --help \n",
    "# option :  -o, --output <file> \n",
    "# Write to file instead of stdout"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1721,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current\n",
      "                                 Dload  Upload   Total   Spent    Left  Speed\n",
      "100 21979  100 21979    0     0  66805      0 --:--:-- --:--:-- --:--:-- 66805\n"
     ]
    }
   ],
   "source": [
    "!curl https://raw.githubusercontent.com/Luc-Bertin/TDs_ESILV/master/td3_discover_pandas/notes_eleves.csv -o \"notes.csv\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1912,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>eleve</th>\n",
       "      <th>note</th>\n",
       "      <th>groupe</th>\n",
       "      <th>quizz</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>eleve0</td>\n",
       "      <td>71,43 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>eleve1</td>\n",
       "      <td>100 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>eleve4</td>\n",
       "      <td>71,43 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>eleve6</td>\n",
       "      <td>42,86 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>eleve8</td>\n",
       "      <td>57,14 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    eleve     note   groupe quizz\n",
       "0  eleve0  71,43 %  Unknown   td1\n",
       "1  eleve1    100 %  Unknown   td1\n",
       "2  eleve4  71,43 %  Unknown   td1\n",
       "3  eleve6  42,86 %  Unknown   td1\n",
       "4  eleve8  57,14 %  Unknown   td1"
      ]
     },
     "execution_count": 1912,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>eleve</th>\n",
       "      <th>note</th>\n",
       "      <th>groupe</th>\n",
       "      <th>quizz</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>741</th>\n",
       "      <td>eleve174</td>\n",
       "      <td>100 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>742</th>\n",
       "      <td>eleve166</td>\n",
       "      <td>66,67 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>743</th>\n",
       "      <td>eleve176</td>\n",
       "      <td>83,33 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>744</th>\n",
       "      <td>eleve186</td>\n",
       "      <td>100 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>745</th>\n",
       "      <td>eleve196</td>\n",
       "      <td>66,67 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        eleve     note groupe quizz\n",
       "741  eleve174    100 %   ibo5   td3\n",
       "742  eleve166  66,67 %   ibo5   td3\n",
       "743  eleve176  83,33 %   ibo5   td3\n",
       "744  eleve186    100 %   ibo5   td3\n",
       "745  eleve196  66,67 %   ibo5   td3"
      ]
     },
     "execution_count": 1912,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes = pd.read_csv(\"notes.csv\", index_col=0)\n",
    "# showing just the first n rows\n",
    "df_notes.head(5)\n",
    "# or the last n rows\n",
    "df_notes.tail(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Let's define a function to be applied for each val in a colum"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1864,
   "metadata": {},
   "outputs": [],
   "source": [
    "def function(val):\n",
    "    \"\"\"A fonction to be applied on each element of a pandas DataFrame column / Series \"\"\"\n",
    "    # we need to return a value for each element computation\n",
    "    return val.upper()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1865,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0        ELEVE0\n",
       "1        ELEVE1\n",
       "2        ELEVE4\n",
       "3        ELEVE6\n",
       "4        ELEVE8\n",
       "         ...   \n",
       "741    ELEVE174\n",
       "742    ELEVE166\n",
       "743    ELEVE176\n",
       "744    ELEVE186\n",
       "745    ELEVE196\n",
       "Name: eleve, Length: 746, dtype: object"
      ]
     },
     "execution_count": 1865,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.eleve.apply(function) # the function applies on each value in the column"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "behind, apply is looping on each element of the column `eleve` and returning a value for each of them"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### We can also use apply on the dataframe "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "But wee need to provide an axis.<br>\n",
    "applied function won't be fed a single column element this time but a Series.<br>\n",
    "a Series whose index depends on the axis we choose !"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1870,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>eleve</th>\n",
       "      <th>note</th>\n",
       "      <th>groupe</th>\n",
       "      <th>quizz</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>eleve0</td>\n",
       "      <td>71,43 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>eleve1</td>\n",
       "      <td>100 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>eleve4</td>\n",
       "      <td>71,43 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>eleve6</td>\n",
       "      <td>42,86 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>eleve8</td>\n",
       "      <td>57,14 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>741</th>\n",
       "      <td>eleve174</td>\n",
       "      <td>100 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>742</th>\n",
       "      <td>eleve166</td>\n",
       "      <td>66,67 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>743</th>\n",
       "      <td>eleve176</td>\n",
       "      <td>83,33 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>744</th>\n",
       "      <td>eleve186</td>\n",
       "      <td>100 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>745</th>\n",
       "      <td>eleve196</td>\n",
       "      <td>66,67 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>746 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        eleve     note   groupe quizz\n",
       "0      eleve0  71,43 %  Unknown   td1\n",
       "1      eleve1    100 %  Unknown   td1\n",
       "2      eleve4  71,43 %  Unknown   td1\n",
       "3      eleve6  42,86 %  Unknown   td1\n",
       "4      eleve8  57,14 %  Unknown   td1\n",
       "..        ...      ...      ...   ...\n",
       "741  eleve174    100 %     ibo5   td3\n",
       "742  eleve166  66,67 %     ibo5   td3\n",
       "743  eleve176  83,33 %     ibo5   td3\n",
       "744  eleve186    100 %     ibo5   td3\n",
       "745  eleve196  66,67 %     ibo5   td3\n",
       "\n",
       "[746 rows x 4 columns]"
      ]
     },
     "execution_count": 1870,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1871,
   "metadata": {},
   "outputs": [],
   "source": [
    "def function(row):\n",
    "    \"\"\"A fonction to be applied on each row of the DataFrame\n",
    "    i.e. each row is indeed a pandas.Series object \n",
    "    passed-in the applied function at each loop iteration.\n",
    "    We will need later on to use axis=1 for Series to be the rows\"\"\"\n",
    "    \n",
    "    # having a full row we can do many things to create\n",
    "    if int( row[\"eleve\"][-1] ) % 2 == 0:\n",
    "        return \"pair\"\n",
    "    return \"impair\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1872,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0        pair\n",
       "1      impair\n",
       "2        pair\n",
       "3        pair\n",
       "4        pair\n",
       "        ...  \n",
       "741      pair\n",
       "742      pair\n",
       "743      pair\n",
       "744      pair\n",
       "745      pair\n",
       "Length: 746, dtype: object"
      ]
     },
     "execution_count": 1872,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.apply(function, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1875,
   "metadata": {},
   "outputs": [],
   "source": [
    "def function(row):\n",
    "    \"\"\"Another function on rows but returning a pandas.Series each time\n",
    "    i.e. then the final result will be a stack of pandas.Series along an axis or the other\n",
    "    i.e. <=> a DataFrame\"\"\"\n",
    "    if int( row[\"eleve\"][-1] ) % 2 == 0:\n",
    "        odd_or_even = \"pair\"\n",
    "    odd_or_even = \"impair\"\n",
    "    \n",
    "    cut_note = row[\"note\"].split(',')[0] # what precedes the comma\n",
    "    return pd.Series([ odd_or_even, cut_note], index=['odd_or_even', 'cut_note'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1876,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>odd_or_even</th>\n",
       "      <th>cut_note</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>impair</td>\n",
       "      <td>71</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>impair</td>\n",
       "      <td>100 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>impair</td>\n",
       "      <td>71</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>impair</td>\n",
       "      <td>42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>impair</td>\n",
       "      <td>57</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>741</th>\n",
       "      <td>impair</td>\n",
       "      <td>100 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>742</th>\n",
       "      <td>impair</td>\n",
       "      <td>66</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>743</th>\n",
       "      <td>impair</td>\n",
       "      <td>83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>744</th>\n",
       "      <td>impair</td>\n",
       "      <td>100 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>745</th>\n",
       "      <td>impair</td>\n",
       "      <td>66</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>746 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    odd_or_even cut_note\n",
       "0        impair       71\n",
       "1        impair    100 %\n",
       "2        impair       71\n",
       "3        impair       42\n",
       "4        impair       57\n",
       "..          ...      ...\n",
       "741      impair    100 %\n",
       "742      impair       66\n",
       "743      impair       83\n",
       "744      impair    100 %\n",
       "745      impair       66\n",
       "\n",
       "[746 rows x 2 columns]"
      ]
     },
     "execution_count": 1876,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.apply(function, axis=1) # a pandas.Series for each row"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To avoid having to define a loop we should use vectorized functions (we will talk about it later on).\n",
    "On integers (not the case here) use of vectorized functions can greatly improve computational speed. (C-loop)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1891,
   "metadata": {},
   "outputs": [],
   "source": [
    "def function(col):\n",
    "    \"\"\"Another function on cols this time\"\"\"\n",
    "    try:\n",
    "        return sum([int(x[:1]) for x in col])\n",
    "    except:\n",
    "        return \"can't sum on this col\"\n",
    "    #return pd.Series([ odd_or_even, cut_note], index=['odd_or_even', 'cut_note'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1892,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "eleve     can't sum on this col\n",
       "note                       3053\n",
       "groupe    can't sum on this col\n",
       "quizz     can't sum on this col\n",
       "dtype: object"
      ]
     },
     "execution_count": 1892,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.apply(function, axis=0) # a pandas.Series for each col"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1898,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3053.0"
      ]
     },
     "execution_count": 1898,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.note.str[:1].astype(float).sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Manipulating columns with strings"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Manipulating columns with strings"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "What is a vectorized function: it is a function that applies on the whole sequence rather than each element as input.\n",
    "\n",
    "This is the case for numpy functions like `np.mean`, `np.sum`, `np.std` which apply on a numerically valued input array as a whole, so the loop is moved from the Python-level to the [C one](https://stackoverflow.com/questions/58071631/is-numpy-vectorization-using-c-loops)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Numeric types include: `int`, `float`, `datetime`, `bool`, `category`. They exclude `object` dtype and can be held in **contiguous memory blocks**. See [here too, concerning C contiguous array stored in memory when creating a numpy array](https://stackoverflow.com/questions/26998223/what-is-the-difference-between-contiguous-and-non-contiguous-arrays).\n",
    "\n",
    "> Why are numpy operations more efficient than simple crude Python ?  as we've seen earlier **Everything in Python is an object**. This includes, unlike C, numbers. Python types therefore have an **overhead which does not exist with native C types**. NumPy methods are **usually C-based.** \n",
    "\n",
    "check [here](https://stackoverflow.com/questions/52673285/performance-of-pandas-apply-vs-np-vectorize-to-create-new-column-from-existing-c)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> np.vectorize is fake vectorisation. According to [documentation]( https://docs.scipy.org/doc/numpy-1.9.1/reference/generated/numpy.vectorize.html): The vectorize function is provided primarily for convenience, not for performance. The implementation is essentially a for loop. It means there is no reazon in vectorize of function wich could be applied directly as it is in your example. Actually this could lead to degraded performance. Main goal of the \"vectorize\" is to hide a for loop from you code. But it will not avoid it neither change expected results.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This link provides a good an example of simple [vectorization](https://engineering.upside.com/a-beginners-guide-to-optimizing-pandas-code-for-speed-c09ef2c6a4d6)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Numpy does not provide vectorization functions for arrays of strings."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Pandas provide vectorized [str operations](https://stackoverflow.com/questions/50744369/how-to-speed-up-pandas-string-function). Pros are that you don't have to write any loop and can take the column/Series as a whole. Cons are that they are not actually faster than using a simply apply. String operations are inherently difficult to vectorize. Pandas treats strings as objects, and all operations on objects fall back to a slow, loopy implementation. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Already provided Pandas vectorized string methods available in `.str.`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1913,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>eleve</th>\n",
       "      <th>note</th>\n",
       "      <th>groupe</th>\n",
       "      <th>quizz</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Eleve0</td>\n",
       "      <td>71,43 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Eleve1</td>\n",
       "      <td>100 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Eleve4</td>\n",
       "      <td>71,43 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Eleve6</td>\n",
       "      <td>42,86 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Eleve8</td>\n",
       "      <td>57,14 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>741</th>\n",
       "      <td>Eleve174</td>\n",
       "      <td>100 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>742</th>\n",
       "      <td>Eleve166</td>\n",
       "      <td>66,67 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>743</th>\n",
       "      <td>Eleve176</td>\n",
       "      <td>83,33 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>744</th>\n",
       "      <td>Eleve186</td>\n",
       "      <td>100 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>745</th>\n",
       "      <td>Eleve196</td>\n",
       "      <td>66,67 %</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>746 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        eleve     note   groupe quizz\n",
       "0      Eleve0  71,43 %  Unknown   td1\n",
       "1      Eleve1    100 %  Unknown   td1\n",
       "2      Eleve4  71,43 %  Unknown   td1\n",
       "3      Eleve6  42,86 %  Unknown   td1\n",
       "4      Eleve8  57,14 %  Unknown   td1\n",
       "..        ...      ...      ...   ...\n",
       "741  Eleve174    100 %     ibo5   td3\n",
       "742  Eleve166  66,67 %     ibo5   td3\n",
       "743  Eleve176  83,33 %     ibo5   td3\n",
       "744  Eleve186    100 %     ibo5   td3\n",
       "745  Eleve196  66,67 %     ibo5   td3\n",
       "\n",
       "[746 rows x 4 columns]"
      ]
     },
     "execution_count": 1913,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes[\"eleve\"] = df_notes.eleve.str.capitalize()\n",
    "df_notes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1724,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0       True\n",
       "1       True\n",
       "2       True\n",
       "3       True\n",
       "4       True\n",
       "       ...  \n",
       "741    False\n",
       "742    False\n",
       "743    False\n",
       "744    False\n",
       "745    False\n",
       "Name: groupe, Length: 746, dtype: bool"
      ]
     },
     "execution_count": 1724,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mask = df_notes.groupe.str.startswith(\"U\")\n",
    "mask"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1725,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>eleve</th>\n",
       "      <th>note</th>\n",
       "      <th>groupe</th>\n",
       "      <th>quizz</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Eleve0</td>\n",
       "      <td>71,43 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Eleve1</td>\n",
       "      <td>100 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Eleve4</td>\n",
       "      <td>71,43 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Eleve6</td>\n",
       "      <td>42,86 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Eleve8</td>\n",
       "      <td>57,14 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>87</th>\n",
       "      <td>Eleve202</td>\n",
       "      <td>57,14 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>88</th>\n",
       "      <td>Eleve203</td>\n",
       "      <td>57,14 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89</th>\n",
       "      <td>Eleve204</td>\n",
       "      <td>71,43 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90</th>\n",
       "      <td>Eleve205</td>\n",
       "      <td>42,86 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>91</th>\n",
       "      <td>Eleve207</td>\n",
       "      <td>42,86 %</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>92 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       eleve     note   groupe quizz\n",
       "0     Eleve0  71,43 %  Unknown   td1\n",
       "1     Eleve1    100 %  Unknown   td1\n",
       "2     Eleve4  71,43 %  Unknown   td1\n",
       "3     Eleve6  42,86 %  Unknown   td1\n",
       "4     Eleve8  57,14 %  Unknown   td1\n",
       "..       ...      ...      ...   ...\n",
       "87  Eleve202  57,14 %  Unknown   td1\n",
       "88  Eleve203  57,14 %  Unknown   td1\n",
       "89  Eleve204  71,43 %  Unknown   td1\n",
       "90  Eleve205  42,86 %  Unknown   td1\n",
       "91  Eleve207  42,86 %  Unknown   td1\n",
       "\n",
       "[92 rows x 4 columns]"
      ]
     },
     "execution_count": 1725,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes[mask]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1726,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0      [71, 43 %]\n",
       "1         [100 %]\n",
       "2      [71, 43 %]\n",
       "3      [42, 86 %]\n",
       "4      [57, 14 %]\n",
       "          ...    \n",
       "741       [100 %]\n",
       "742    [66, 67 %]\n",
       "743    [83, 33 %]\n",
       "744       [100 %]\n",
       "745    [66, 67 %]\n",
       "Name: note, Length: 746, dtype: object"
      ]
     },
     "execution_count": 1726,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.note.str.split(',')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1727,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0       71.43\n",
       "1      100.00\n",
       "2       71.43\n",
       "3       42.86\n",
       "4       57.14\n",
       "        ...  \n",
       "741    100.00\n",
       "742     66.67\n",
       "743     83.33\n",
       "744    100.00\n",
       "745     66.67\n",
       "Name: note, Length: 746, dtype: float64"
      ]
     },
     "execution_count": 1727,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(df_notes.note\n",
    " .str.replace(\"%\",\"\") # replace all occurences of \"%\" as \"\"\n",
    " .str.replace(\",\", \".\") # replace all occurences of \",\" as \".\"\n",
    " .astype(float)\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1728,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0       71.43\n",
       "1      100.00\n",
       "2       71.43\n",
       "3       42.86\n",
       "4       57.14\n",
       "        ...  \n",
       "741    100.00\n",
       "742     66.67\n",
       "743     83.33\n",
       "744    100.00\n",
       "745     66.67\n",
       "Name: note, Length: 746, dtype: float64"
      ]
     },
     "execution_count": 1728,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(df_notes.note\n",
    " .str.findall(\"(\\d+),?(\\d+)?\") #regex to find all matching groups in each element of the Series\n",
    " .str[0] # vectorized element access in the column, works for all iterable, hence even a list in a pd.Series, \n",
    " .str.join(\".\") # join the lists with \".\" rather than ','\n",
    " .str.rstrip('.') # take off the last dot if exists\n",
    " .astype(float) # convert to float type\n",
    ") "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1914,
   "metadata": {},
   "outputs": [],
   "source": [
    "serie_notes =\\\n",
    "( \n",
    "df_notes.note\n",
    " .str.extract(\"(\\d+),?(\\d+)?\") # expand to multiple cols\n",
    " .fillna(0) # fill NaN as 0 when no matched group\n",
    " .astype(float) # convert to float\n",
    ")\n",
    "serie_notes[0] += serie_notes[1]/100\n",
    "serie_notes.drop(1, axis=1,inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1915,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_notes.note = serie_notes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1916,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>eleve</th>\n",
       "      <th>note</th>\n",
       "      <th>groupe</th>\n",
       "      <th>quizz</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Eleve0</td>\n",
       "      <td>71.43</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Eleve1</td>\n",
       "      <td>100.00</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Eleve4</td>\n",
       "      <td>71.43</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Eleve6</td>\n",
       "      <td>42.86</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Eleve8</td>\n",
       "      <td>57.14</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>741</th>\n",
       "      <td>Eleve174</td>\n",
       "      <td>100.00</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>742</th>\n",
       "      <td>Eleve166</td>\n",
       "      <td>66.67</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>743</th>\n",
       "      <td>Eleve176</td>\n",
       "      <td>83.33</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>744</th>\n",
       "      <td>Eleve186</td>\n",
       "      <td>100.00</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>745</th>\n",
       "      <td>Eleve196</td>\n",
       "      <td>66.67</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>746 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        eleve    note   groupe quizz\n",
       "0      Eleve0   71.43  Unknown   td1\n",
       "1      Eleve1  100.00  Unknown   td1\n",
       "2      Eleve4   71.43  Unknown   td1\n",
       "3      Eleve6   42.86  Unknown   td1\n",
       "4      Eleve8   57.14  Unknown   td1\n",
       "..        ...     ...      ...   ...\n",
       "741  Eleve174  100.00     ibo5   td3\n",
       "742  Eleve166   66.67     ibo5   td3\n",
       "743  Eleve176   83.33     ibo5   td3\n",
       "744  Eleve186  100.00     ibo5   td3\n",
       "745  Eleve196   66.67     ibo5   td3\n",
       "\n",
       "[746 rows x 4 columns]"
      ]
     },
     "execution_count": 1916,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Other interesting functions to mention"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To compute the counts of unique values use : pd.Series.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1918,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ibo1       117\n",
       "ibo5       115\n",
       "ibo7       114\n",
       "Unknown     92\n",
       "ibo6        85\n",
       "ibo3        85\n",
       "ibo4        81\n",
       "ibo2        57\n",
       "Name: groupe, dtype: int64"
      ]
     },
     "execution_count": 1918,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.groupe.value_counts(ascending=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To do a `binning` : i.e. group a number of more or less continuous values into a smaller number of \"bins\". Use pd.cut"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1924,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0       (60.0, 80.0]\n",
       "1      (80.0, 100.0]\n",
       "2       (60.0, 80.0]\n",
       "3       (40.0, 60.0]\n",
       "4       (40.0, 60.0]\n",
       "           ...      \n",
       "741    (80.0, 100.0]\n",
       "742     (60.0, 80.0]\n",
       "743    (80.0, 100.0]\n",
       "744    (80.0, 100.0]\n",
       "745     (60.0, 80.0]\n",
       "Name: note, Length: 746, dtype: category\n",
       "Categories (5, interval[float64]): [(-0.1, 20.0] < (20.0, 40.0] < (40.0, 60.0] < (60.0, 80.0] < (80.0, 100.0]]"
      ]
     },
     "execution_count": 1924,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.cut(df_notes.note, bins=5) # 5 equal sized bins"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1927,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0       (50, 75]\n",
       "1      (75, 100]\n",
       "2       (50, 75]\n",
       "3        (0, 50]\n",
       "4       (50, 75]\n",
       "         ...    \n",
       "741    (75, 100]\n",
       "742     (50, 75]\n",
       "743    (75, 100]\n",
       "744    (75, 100]\n",
       "745     (50, 75]\n",
       "Name: note, Length: 746, dtype: category\n",
       "Categories (3, interval[int64]): [(0, 50] < (50, 75] < (75, 100]]"
      ]
     },
     "execution_count": 1927,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.cut(df_notes.note, bins=[0, 50, 75, 100])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1929,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Bin labels must be one fewer than the number of bin edges\n"
     ]
    }
   ],
   "source": [
    "try: \n",
    "    pd.cut(df_notes.note, bins=[0, 50, 75, 100], labels=[\"Bad\"])\n",
    "except Exception as e:\n",
    "    print(e)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1932,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0        Ok\n",
       "1      Good\n",
       "2        Ok\n",
       "3       Bad\n",
       "4        Ok\n",
       "       ... \n",
       "741    Good\n",
       "742      Ok\n",
       "743    Good\n",
       "744    Good\n",
       "745      Ok\n",
       "Name: note, Length: 746, dtype: category\n",
       "Categories (4, object): [Very Bad < Bad < Ok < Good]"
      ]
     },
     "execution_count": 1932,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes['appreciation'] = pd.cut(df_notes.note, bins=[0, 25, 50, 75, 100], labels=[\"Very Bad\", \"Bad\", \"Ok\", \"Good\"])\n",
    "df_notes.appreciation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## GroupBy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1732,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>eleve</th>\n",
       "      <th>note</th>\n",
       "      <th>groupe</th>\n",
       "      <th>quizz</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Eleve0</td>\n",
       "      <td>71.43</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Eleve1</td>\n",
       "      <td>100.00</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Eleve4</td>\n",
       "      <td>71.43</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    eleve    note   groupe quizz\n",
       "0  Eleve0   71.43  Unknown   td1\n",
       "1  Eleve1  100.00  Unknown   td1\n",
       "2  Eleve4   71.43  Unknown   td1"
      ]
     },
     "execution_count": 1732,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Groupby applies the ***\"split, apply, combine\"*** method."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- We first have to use a `key` to groupby, i.e. a column of different labels that will serve to ***split*** the main `df` into different subsets (one for each label in the concerned column), just as we would do a GROUP BY in SQL syntax."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1621,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<pandas.core.groupby.generic.DataFrameGroupBy object at 0x1187d9a90>"
      ]
     },
     "execution_count": 1621,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.groupby('groupe') "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "No computation is done yet"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This result in a DataFrameGroupBy object we can iterate on."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1625,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "label used Unknown, dataframe shape (92, 4)\n",
      "label used ibo1, dataframe shape (117, 4)\n",
      "label used ibo2, dataframe shape (57, 4)\n",
      "label used ibo3, dataframe shape (85, 4)\n",
      "label used ibo4, dataframe shape (81, 4)\n",
      "label used ibo5, dataframe shape (115, 4)\n",
      "label used ibo6, dataframe shape (85, 4)\n",
      "label used ibo7, dataframe shape (114, 4)\n"
     ]
    }
   ],
   "source": [
    "for name_group, group in df_notes.groupby('groupe'):\n",
    "    # the label used , the df subset (one for each label)\n",
    "    print( \"label used {}, dataframe shape {}\".format(name_group,group.shape)) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Notice that we are not limited by grouping over one column keys."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1628,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "label used ('Unknown', 'td1'), dataframe shape (92, 4)\n",
      "label used ('ibo1', 'td1'), dataframe shape (30, 4)\n",
      "label used ('ibo1', 'td2'), dataframe shape (30, 4)\n",
      "label used ('ibo1', 'td3'), dataframe shape (27, 4)\n",
      "label used ('ibo1', 'td4'), dataframe shape (30, 4)\n",
      "label used ('ibo2', 'td2'), dataframe shape (30, 4)\n",
      "label used ('ibo2', 'td3'), dataframe shape (27, 4)\n",
      "label used ('ibo3', 'td2'), dataframe shape (30, 4)\n",
      "label used ('ibo3', 'td3'), dataframe shape (27, 4)\n",
      "label used ('ibo3', 'td4'), dataframe shape (28, 4)\n",
      "label used ('ibo4', 'td2'), dataframe shape (27, 4)\n",
      "label used ('ibo4', 'td3'), dataframe shape (28, 4)\n",
      "label used ('ibo4', 'td4'), dataframe shape (26, 4)\n",
      "label used ('ibo5', 'td1'), dataframe shape (27, 4)\n",
      "label used ('ibo5', 'td2'), dataframe shape (30, 4)\n",
      "label used ('ibo5', 'td3'), dataframe shape (28, 4)\n",
      "label used ('ibo5', 'td4'), dataframe shape (30, 4)\n",
      "label used ('ibo6', 'td2'), dataframe shape (29, 4)\n",
      "label used ('ibo6', 'td3'), dataframe shape (28, 4)\n",
      "label used ('ibo6', 'td4'), dataframe shape (28, 4)\n",
      "label used ('ibo7', 'td1'), dataframe shape (29, 4)\n",
      "label used ('ibo7', 'td2'), dataframe shape (28, 4)\n",
      "label used ('ibo7', 'td3'), dataframe shape (28, 4)\n",
      "label used ('ibo7', 'td4'), dataframe shape (29, 4)\n"
     ]
    }
   ],
   "source": [
    "for name_group, group in df_notes.groupby(['groupe', \"quizz\"]):\n",
    "    # the label used , the df subset (one for each label)\n",
    "    print( \"label used {}, dataframe shape {}\".format(name_group,group.shape)) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This results in a mutli-index with:\n",
    "- level0 = the group\n",
    "- level1 = the quizz number"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can also index the `GroupByDataFrame` object by retrieving one Series (again no computation is done yet)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1637,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<pandas.core.groupby.generic.SeriesGroupBy object at 0x1188bdeb8>"
      ]
     },
     "execution_count": 1637,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.groupby(['groupe', \"quizz\"])[\"note\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1638,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "label used ('Unknown', 'td1'), \n",
      "<class 'pandas.core.series.Series'> shape (92,)\n",
      "label used ('ibo1', 'td1'), \n",
      "<class 'pandas.core.series.Series'> shape (30,)\n",
      "label used ('ibo1', 'td2'), \n",
      "<class 'pandas.core.series.Series'> shape (30,)\n",
      "label used ('ibo1', 'td3'), \n",
      "<class 'pandas.core.series.Series'> shape (27,)\n",
      "label used ('ibo1', 'td4'), \n",
      "<class 'pandas.core.series.Series'> shape (30,)\n",
      "label used ('ibo2', 'td2'), \n",
      "<class 'pandas.core.series.Series'> shape (30,)\n",
      "label used ('ibo2', 'td3'), \n",
      "<class 'pandas.core.series.Series'> shape (27,)\n",
      "label used ('ibo3', 'td2'), \n",
      "<class 'pandas.core.series.Series'> shape (30,)\n",
      "label used ('ibo3', 'td3'), \n",
      "<class 'pandas.core.series.Series'> shape (27,)\n",
      "label used ('ibo3', 'td4'), \n",
      "<class 'pandas.core.series.Series'> shape (28,)\n",
      "label used ('ibo4', 'td2'), \n",
      "<class 'pandas.core.series.Series'> shape (27,)\n",
      "label used ('ibo4', 'td3'), \n",
      "<class 'pandas.core.series.Series'> shape (28,)\n",
      "label used ('ibo4', 'td4'), \n",
      "<class 'pandas.core.series.Series'> shape (26,)\n",
      "label used ('ibo5', 'td1'), \n",
      "<class 'pandas.core.series.Series'> shape (27,)\n",
      "label used ('ibo5', 'td2'), \n",
      "<class 'pandas.core.series.Series'> shape (30,)\n",
      "label used ('ibo5', 'td3'), \n",
      "<class 'pandas.core.series.Series'> shape (28,)\n",
      "label used ('ibo5', 'td4'), \n",
      "<class 'pandas.core.series.Series'> shape (30,)\n",
      "label used ('ibo6', 'td2'), \n",
      "<class 'pandas.core.series.Series'> shape (29,)\n",
      "label used ('ibo6', 'td3'), \n",
      "<class 'pandas.core.series.Series'> shape (28,)\n",
      "label used ('ibo6', 'td4'), \n",
      "<class 'pandas.core.series.Series'> shape (28,)\n",
      "label used ('ibo7', 'td1'), \n",
      "<class 'pandas.core.series.Series'> shape (29,)\n",
      "label used ('ibo7', 'td2'), \n",
      "<class 'pandas.core.series.Series'> shape (28,)\n",
      "label used ('ibo7', 'td3'), \n",
      "<class 'pandas.core.series.Series'> shape (28,)\n",
      "label used ('ibo7', 'td4'), \n",
      "<class 'pandas.core.series.Series'> shape (29,)\n"
     ]
    }
   ],
   "source": [
    "for name_group, group in df_notes.groupby(['groupe', \"quizz\"])[\"note\"]:\n",
    "    print( \"label used {}, \\n{} shape {}\".format(name_group, type(group), group.shape)) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Aggregation functions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now think about the \"apply, combine\" part"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1733,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "eleve      object\n",
       "note      float64\n",
       "groupe     object\n",
       "quizz      object\n",
       "dtype: object"
      ]
     },
     "execution_count": 1733,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.dtypes "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Pandas provides us some functions to be applied on a dataframe or Series (.mean(), .sum(), .std(), .describe(), .min(), etc...), we can seemlessly append one of them to the GroupBy Object to operate **on each of the subsets DataFrames/Series created on the split step** (this is the ***apply*** step).\n",
    "\n",
    "After applying the function to each split, a ***combined*** result is returned, in the form of a Series object or DataFrame.\n",
    "\n",
    "Note that for those aggregating functions reduce the shape of the data e.g. summing or meaning on a Series result in a scalar (the sum or the mean), this will be operated over each Series groups from the **split** step."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1734,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "groupe\n",
       "Unknown    63.664022\n",
       "ibo1       87.337607\n",
       "ibo2       97.251053\n",
       "ibo3       86.418824\n",
       "ibo4       87.953580\n",
       "ibo5       80.288957\n",
       "ibo6       83.484000\n",
       "ibo7       86.402456\n",
       "Name: note, dtype: float64"
      ]
     },
     "execution_count": 1734,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.groupby(['groupe'])[\"note\"].mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1737,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "groupe   quizz\n",
       "Unknown  td1      63.664022\n",
       "ibo1     td1      67.618000\n",
       "         td2      94.666667\n",
       "         td3      90.739630\n",
       "         td4      96.666333\n",
       "ibo2     td2      98.666667\n",
       "         td3      95.678148\n",
       "ibo3     td2      88.666667\n",
       "         td3      88.887037\n",
       "         td4      81.630357\n",
       "ibo4     td2      87.407407\n",
       "         td3      87.500000\n",
       "         td4      89.009231\n",
       "ibo5     td1      53.967407\n",
       "         td2      90.000000\n",
       "         td3      90.475714\n",
       "         td4      84.759667\n",
       "ibo6     td2      90.344828\n",
       "         td3      83.332857\n",
       "         td4      76.529286\n",
       "ibo7     td1      74.875517\n",
       "         td2      85.714286\n",
       "         td3      94.642500\n",
       "         td4      90.637931\n",
       "Name: note, dtype: float64"
      ]
     },
     "execution_count": 1737,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.groupby(['groupe', 'quizz'])[\"note\"].mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "as we get a hierarchical index we can unstack to make use of the dimensionality brought by column indexesm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1738,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>quizz</th>\n",
       "      <th>td1</th>\n",
       "      <th>td2</th>\n",
       "      <th>td3</th>\n",
       "      <th>td4</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>groupe</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Unknown</th>\n",
       "      <td>63.664022</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo1</th>\n",
       "      <td>67.618000</td>\n",
       "      <td>94.666667</td>\n",
       "      <td>90.739630</td>\n",
       "      <td>96.666333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo2</th>\n",
       "      <td>NaN</td>\n",
       "      <td>98.666667</td>\n",
       "      <td>95.678148</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo3</th>\n",
       "      <td>NaN</td>\n",
       "      <td>88.666667</td>\n",
       "      <td>88.887037</td>\n",
       "      <td>81.630357</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo4</th>\n",
       "      <td>NaN</td>\n",
       "      <td>87.407407</td>\n",
       "      <td>87.500000</td>\n",
       "      <td>89.009231</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo5</th>\n",
       "      <td>53.967407</td>\n",
       "      <td>90.000000</td>\n",
       "      <td>90.475714</td>\n",
       "      <td>84.759667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo6</th>\n",
       "      <td>NaN</td>\n",
       "      <td>90.344828</td>\n",
       "      <td>83.332857</td>\n",
       "      <td>76.529286</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo7</th>\n",
       "      <td>74.875517</td>\n",
       "      <td>85.714286</td>\n",
       "      <td>94.642500</td>\n",
       "      <td>90.637931</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "quizz          td1        td2        td3        td4\n",
       "groupe                                             \n",
       "Unknown  63.664022        NaN        NaN        NaN\n",
       "ibo1     67.618000  94.666667  90.739630  96.666333\n",
       "ibo2           NaN  98.666667  95.678148        NaN\n",
       "ibo3           NaN  88.666667  88.887037  81.630357\n",
       "ibo4           NaN  87.407407  87.500000  89.009231\n",
       "ibo5     53.967407  90.000000  90.475714  84.759667\n",
       "ibo6           NaN  90.344828  83.332857  76.529286\n",
       "ibo7     74.875517  85.714286  94.642500  90.637931"
      ]
     },
     "execution_count": 1738,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "_.unstack()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Something is unusual? why is there `NaN`? some class groups should have grades for each quizz."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1739,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "eleve     0\n",
       "note      0\n",
       "groupe    0\n",
       "quizz     0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1739,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "though all the data seems complete..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1711,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "eleve     0\n",
       "note      0\n",
       "groupe    0\n",
       "quizz     0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 1711,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.isnull().apply(sum, axis=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Notice the `Unknown` group, we should look more into this..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1934,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>eleve</th>\n",
       "      <th>note</th>\n",
       "      <th>groupe</th>\n",
       "      <th>quizz</th>\n",
       "      <th>appreciation</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Eleve0</td>\n",
       "      <td>71.43</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "      <td>Ok</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Eleve1</td>\n",
       "      <td>100.00</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "      <td>Good</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Eleve4</td>\n",
       "      <td>71.43</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "      <td>Ok</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Eleve6</td>\n",
       "      <td>42.86</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "      <td>Bad</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Eleve8</td>\n",
       "      <td>57.14</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "      <td>Ok</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>87</th>\n",
       "      <td>Eleve202</td>\n",
       "      <td>57.14</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "      <td>Ok</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>88</th>\n",
       "      <td>Eleve203</td>\n",
       "      <td>57.14</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "      <td>Ok</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89</th>\n",
       "      <td>Eleve204</td>\n",
       "      <td>71.43</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "      <td>Ok</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90</th>\n",
       "      <td>Eleve205</td>\n",
       "      <td>42.86</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "      <td>Bad</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>91</th>\n",
       "      <td>Eleve207</td>\n",
       "      <td>42.86</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>td1</td>\n",
       "      <td>Bad</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>92 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       eleve    note   groupe quizz appreciation\n",
       "0     Eleve0   71.43  Unknown   td1           Ok\n",
       "1     Eleve1  100.00  Unknown   td1         Good\n",
       "2     Eleve4   71.43  Unknown   td1           Ok\n",
       "3     Eleve6   42.86  Unknown   td1          Bad\n",
       "4     Eleve8   57.14  Unknown   td1           Ok\n",
       "..       ...     ...      ...   ...          ...\n",
       "87  Eleve202   57.14  Unknown   td1           Ok\n",
       "88  Eleve203   57.14  Unknown   td1           Ok\n",
       "89  Eleve204   71.43  Unknown   td1           Ok\n",
       "90  Eleve205   42.86  Unknown   td1          Bad\n",
       "91  Eleve207   42.86  Unknown   td1          Bad\n",
       "\n",
       "[92 rows x 5 columns]"
      ]
     },
     "execution_count": 1934,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes[df_notes.groupe == \"Unknown\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1763,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "92"
      ]
     },
     "execution_count": 1763,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "_.shape[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To apply multiple aggregate functions at once using a list of the functions you want to apply in aggregate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1749,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr:last-of-type th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"2\" halign=\"left\">note</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>max</th>\n",
       "      <th>min</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>quizz</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>td1</th>\n",
       "      <td>100.0</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>td2</th>\n",
       "      <td>100.0</td>\n",
       "      <td>40.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>td3</th>\n",
       "      <td>100.0</td>\n",
       "      <td>50.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>td4</th>\n",
       "      <td>100.0</td>\n",
       "      <td>42.86</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        note       \n",
       "         max    min\n",
       "quizz              \n",
       "td1    100.0   0.00\n",
       "td2    100.0  40.00\n",
       "td3    100.0  50.00\n",
       "td4    100.0  42.86"
      ]
     },
     "execution_count": 1749,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.groupby('quizz').agg({'note': ['max', min]})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "this results in a multi column index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1744,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MultiIndex([('note', 'max'),\n",
       "            ('note', 'min')],\n",
       "           )"
      ]
     },
     "execution_count": 1744,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "_.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Grouping by students, we may have more insight."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1765,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>note</th>\n",
       "      <th>groupe</th>\n",
       "      <th>quizz</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>eleve</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Eleve0</th>\n",
       "      <td>[71.43, 80.0, 100.0]</td>\n",
       "      <td>[Unknown, ibo2, ibo2]</td>\n",
       "      <td>[td1, td2, td3]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve1</th>\n",
       "      <td>[100.0, 100.0, 100.0]</td>\n",
       "      <td>[Unknown, ibo2, ibo2]</td>\n",
       "      <td>[td1, td2, td3]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve10</th>\n",
       "      <td>[100.0, 100.0, 85.71, 71.43]</td>\n",
       "      <td>[ibo7, ibo7, ibo7, ibo7]</td>\n",
       "      <td>[td3, td2, td4, td1]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve100</th>\n",
       "      <td>[100.0, 100.0, 100.0, 85.71]</td>\n",
       "      <td>[ibo7, ibo7, ibo7, ibo7]</td>\n",
       "      <td>[td3, td2, td4, td1]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve101</th>\n",
       "      <td>[100.0, 100.0, 85.71, 100.0]</td>\n",
       "      <td>[ibo7, ibo7, ibo7, ibo7]</td>\n",
       "      <td>[td3, td2, td4, td1]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve95</th>\n",
       "      <td>[85.71, 100.0, 100.0]</td>\n",
       "      <td>[Unknown, ibo2, ibo2]</td>\n",
       "      <td>[td1, td2, td3]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve96</th>\n",
       "      <td>[100.0, 100.0, 100.0, 42.86]</td>\n",
       "      <td>[ibo1, ibo1, ibo1, ibo1]</td>\n",
       "      <td>[td4, td3, td2, td1]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve97</th>\n",
       "      <td>[28.57, 85.71, 83.33, 60.0]</td>\n",
       "      <td>[Unknown, ibo4, ibo4, ibo4]</td>\n",
       "      <td>[td1, td4, td3, td2]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve98</th>\n",
       "      <td>[42.86, 85.71, 100.0, 100.0]</td>\n",
       "      <td>[Unknown, ibo4, ibo4, ibo4]</td>\n",
       "      <td>[td1, td4, td3, td2]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve99</th>\n",
       "      <td>[71.43, 100.0, 100.0]</td>\n",
       "      <td>[Unknown, ibo2, ibo2]</td>\n",
       "      <td>[td1, td2, td3]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>208 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                  note                       groupe  \\\n",
       "eleve                                                                 \n",
       "Eleve0            [71.43, 80.0, 100.0]        [Unknown, ibo2, ibo2]   \n",
       "Eleve1           [100.0, 100.0, 100.0]        [Unknown, ibo2, ibo2]   \n",
       "Eleve10   [100.0, 100.0, 85.71, 71.43]     [ibo7, ibo7, ibo7, ibo7]   \n",
       "Eleve100  [100.0, 100.0, 100.0, 85.71]     [ibo7, ibo7, ibo7, ibo7]   \n",
       "Eleve101  [100.0, 100.0, 85.71, 100.0]     [ibo7, ibo7, ibo7, ibo7]   \n",
       "...                                ...                          ...   \n",
       "Eleve95          [85.71, 100.0, 100.0]        [Unknown, ibo2, ibo2]   \n",
       "Eleve96   [100.0, 100.0, 100.0, 42.86]     [ibo1, ibo1, ibo1, ibo1]   \n",
       "Eleve97    [28.57, 85.71, 83.33, 60.0]  [Unknown, ibo4, ibo4, ibo4]   \n",
       "Eleve98   [42.86, 85.71, 100.0, 100.0]  [Unknown, ibo4, ibo4, ibo4]   \n",
       "Eleve99          [71.43, 100.0, 100.0]        [Unknown, ibo2, ibo2]   \n",
       "\n",
       "                         quizz  \n",
       "eleve                           \n",
       "Eleve0         [td1, td2, td3]  \n",
       "Eleve1         [td1, td2, td3]  \n",
       "Eleve10   [td3, td2, td4, td1]  \n",
       "Eleve100  [td3, td2, td4, td1]  \n",
       "Eleve101  [td3, td2, td4, td1]  \n",
       "...                        ...  \n",
       "Eleve95        [td1, td2, td3]  \n",
       "Eleve96   [td4, td3, td2, td1]  \n",
       "Eleve97   [td1, td4, td3, td2]  \n",
       "Eleve98   [td1, td4, td3, td2]  \n",
       "Eleve99        [td1, td2, td3]  \n",
       "\n",
       "[208 rows x 3 columns]"
      ]
     },
     "execution_count": 1765,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.groupby('eleve').agg(list)\n",
    "# applied on all columns (where the function can be used on) for each subset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Some students are known, but are not always written as such."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### transform"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "sometimes we want the ***\"apply-combine\"*** steps to avoid reducing the data size but compute for each data record something based on some intra-group/splits caracteristics "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here we would want for example to group by students and replace `Unknown` fields by the correct information from the other students record."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1783,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>eleve</th>\n",
       "      <th>note</th>\n",
       "      <th>groupe</th>\n",
       "      <th>quizz</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Eleve0</td>\n",
       "      <td>71.43</td>\n",
       "      <td>ibo2</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Eleve1</td>\n",
       "      <td>100.00</td>\n",
       "      <td>ibo2</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Eleve4</td>\n",
       "      <td>71.43</td>\n",
       "      <td>ibo2</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Eleve6</td>\n",
       "      <td>42.86</td>\n",
       "      <td>ibo6</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Eleve8</td>\n",
       "      <td>57.14</td>\n",
       "      <td>ibo4</td>\n",
       "      <td>td1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>741</th>\n",
       "      <td>Eleve174</td>\n",
       "      <td>100.00</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>742</th>\n",
       "      <td>Eleve166</td>\n",
       "      <td>66.67</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>743</th>\n",
       "      <td>Eleve176</td>\n",
       "      <td>83.33</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>744</th>\n",
       "      <td>Eleve186</td>\n",
       "      <td>100.00</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>745</th>\n",
       "      <td>Eleve196</td>\n",
       "      <td>66.67</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>746 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        eleve    note groupe quizz\n",
       "0      Eleve0   71.43   ibo2   td1\n",
       "1      Eleve1  100.00   ibo2   td1\n",
       "2      Eleve4   71.43   ibo2   td1\n",
       "3      Eleve6   42.86   ibo6   td1\n",
       "4      Eleve8   57.14   ibo4   td1\n",
       "..        ...     ...    ...   ...\n",
       "741  Eleve174  100.00   ibo5   td3\n",
       "742  Eleve166   66.67   ibo5   td3\n",
       "743  Eleve176   83.33   ibo5   td3\n",
       "744  Eleve186  100.00   ibo5   td3\n",
       "745  Eleve196   66.67   ibo5   td3\n",
       "\n",
       "[746 rows x 4 columns]"
      ]
     },
     "execution_count": 1783,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1935,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_notes.replace({\"Unknown\":np.nan}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1936,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_notes[\"groupe\"] = df_notes.groupby('eleve')['groupe'].transform(lambda x: x.bfill().ffill())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1937,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>note</th>\n",
       "      <th>groupe</th>\n",
       "      <th>quizz</th>\n",
       "      <th>appreciation</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>eleve</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Eleve0</th>\n",
       "      <td>[71.43, 80.0, 100.0]</td>\n",
       "      <td>[ibo2, ibo2, ibo2]</td>\n",
       "      <td>[td1, td2, td3]</td>\n",
       "      <td>[Ok, Good, Good]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve1</th>\n",
       "      <td>[100.0, 100.0, 100.0]</td>\n",
       "      <td>[ibo2, ibo2, ibo2]</td>\n",
       "      <td>[td1, td2, td3]</td>\n",
       "      <td>[Good, Good, Good]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve10</th>\n",
       "      <td>[100.0, 100.0, 85.71, 71.43]</td>\n",
       "      <td>[ibo7, ibo7, ibo7, ibo7]</td>\n",
       "      <td>[td3, td2, td4, td1]</td>\n",
       "      <td>[Good, Good, Good, Ok]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve100</th>\n",
       "      <td>[100.0, 100.0, 100.0, 85.71]</td>\n",
       "      <td>[ibo7, ibo7, ibo7, ibo7]</td>\n",
       "      <td>[td3, td2, td4, td1]</td>\n",
       "      <td>[Good, Good, Good, Good]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve101</th>\n",
       "      <td>[100.0, 100.0, 85.71, 100.0]</td>\n",
       "      <td>[ibo7, ibo7, ibo7, ibo7]</td>\n",
       "      <td>[td3, td2, td4, td1]</td>\n",
       "      <td>[Good, Good, Good, Good]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve95</th>\n",
       "      <td>[85.71, 100.0, 100.0]</td>\n",
       "      <td>[ibo2, ibo2, ibo2]</td>\n",
       "      <td>[td1, td2, td3]</td>\n",
       "      <td>[Good, Good, Good]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve96</th>\n",
       "      <td>[100.0, 100.0, 100.0, 42.86]</td>\n",
       "      <td>[ibo1, ibo1, ibo1, ibo1]</td>\n",
       "      <td>[td4, td3, td2, td1]</td>\n",
       "      <td>[Good, Good, Good, Bad]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve97</th>\n",
       "      <td>[28.57, 85.71, 83.33, 60.0]</td>\n",
       "      <td>[ibo4, ibo4, ibo4, ibo4]</td>\n",
       "      <td>[td1, td4, td3, td2]</td>\n",
       "      <td>[Bad, Good, Good, Ok]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve98</th>\n",
       "      <td>[42.86, 85.71, 100.0, 100.0]</td>\n",
       "      <td>[ibo4, ibo4, ibo4, ibo4]</td>\n",
       "      <td>[td1, td4, td3, td2]</td>\n",
       "      <td>[Bad, Good, Good, Good]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Eleve99</th>\n",
       "      <td>[71.43, 100.0, 100.0]</td>\n",
       "      <td>[ibo2, ibo2, ibo2]</td>\n",
       "      <td>[td1, td2, td3]</td>\n",
       "      <td>[Ok, Good, Good]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>208 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                  note                    groupe  \\\n",
       "eleve                                                              \n",
       "Eleve0            [71.43, 80.0, 100.0]        [ibo2, ibo2, ibo2]   \n",
       "Eleve1           [100.0, 100.0, 100.0]        [ibo2, ibo2, ibo2]   \n",
       "Eleve10   [100.0, 100.0, 85.71, 71.43]  [ibo7, ibo7, ibo7, ibo7]   \n",
       "Eleve100  [100.0, 100.0, 100.0, 85.71]  [ibo7, ibo7, ibo7, ibo7]   \n",
       "Eleve101  [100.0, 100.0, 85.71, 100.0]  [ibo7, ibo7, ibo7, ibo7]   \n",
       "...                                ...                       ...   \n",
       "Eleve95          [85.71, 100.0, 100.0]        [ibo2, ibo2, ibo2]   \n",
       "Eleve96   [100.0, 100.0, 100.0, 42.86]  [ibo1, ibo1, ibo1, ibo1]   \n",
       "Eleve97    [28.57, 85.71, 83.33, 60.0]  [ibo4, ibo4, ibo4, ibo4]   \n",
       "Eleve98   [42.86, 85.71, 100.0, 100.0]  [ibo4, ibo4, ibo4, ibo4]   \n",
       "Eleve99          [71.43, 100.0, 100.0]        [ibo2, ibo2, ibo2]   \n",
       "\n",
       "                         quizz              appreciation  \n",
       "eleve                                                     \n",
       "Eleve0         [td1, td2, td3]          [Ok, Good, Good]  \n",
       "Eleve1         [td1, td2, td3]        [Good, Good, Good]  \n",
       "Eleve10   [td3, td2, td4, td1]    [Good, Good, Good, Ok]  \n",
       "Eleve100  [td3, td2, td4, td1]  [Good, Good, Good, Good]  \n",
       "Eleve101  [td3, td2, td4, td1]  [Good, Good, Good, Good]  \n",
       "...                        ...                       ...  \n",
       "Eleve95        [td1, td2, td3]        [Good, Good, Good]  \n",
       "Eleve96   [td4, td3, td2, td1]   [Good, Good, Good, Bad]  \n",
       "Eleve97   [td1, td4, td3, td2]     [Bad, Good, Good, Ok]  \n",
       "Eleve98   [td1, td4, td3, td2]   [Bad, Good, Good, Good]  \n",
       "Eleve99        [td1, td2, td3]          [Ok, Good, Good]  \n",
       "\n",
       "[208 rows x 4 columns]"
      ]
     },
     "execution_count": 1937,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.groupby('eleve').agg(list)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's check if we still have some NAs?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1785,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>quizz</th>\n",
       "      <th>td1</th>\n",
       "      <th>td2</th>\n",
       "      <th>td3</th>\n",
       "      <th>td4</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>groupe</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>ibo1</th>\n",
       "      <td>67.618000</td>\n",
       "      <td>94.666667</td>\n",
       "      <td>90.739630</td>\n",
       "      <td>96.666333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo2</th>\n",
       "      <td>76.846207</td>\n",
       "      <td>98.666667</td>\n",
       "      <td>95.678148</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo3</th>\n",
       "      <td>40.002000</td>\n",
       "      <td>88.666667</td>\n",
       "      <td>88.887037</td>\n",
       "      <td>81.630357</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo4</th>\n",
       "      <td>56.632500</td>\n",
       "      <td>87.407407</td>\n",
       "      <td>87.500000</td>\n",
       "      <td>89.009231</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo5</th>\n",
       "      <td>53.967407</td>\n",
       "      <td>90.000000</td>\n",
       "      <td>90.475714</td>\n",
       "      <td>84.759667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo6</th>\n",
       "      <td>61.427667</td>\n",
       "      <td>90.344828</td>\n",
       "      <td>83.332857</td>\n",
       "      <td>76.529286</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo7</th>\n",
       "      <td>74.875517</td>\n",
       "      <td>85.714286</td>\n",
       "      <td>94.642500</td>\n",
       "      <td>90.637931</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "quizz         td1        td2        td3        td4\n",
       "groupe                                            \n",
       "ibo1    67.618000  94.666667  90.739630  96.666333\n",
       "ibo2    76.846207  98.666667  95.678148        NaN\n",
       "ibo3    40.002000  88.666667  88.887037  81.630357\n",
       "ibo4    56.632500  87.407407  87.500000  89.009231\n",
       "ibo5    53.967407  90.000000  90.475714  84.759667\n",
       "ibo6    61.427667  90.344828  83.332857  76.529286\n",
       "ibo7    74.875517  85.714286  94.642500  90.637931"
      ]
     },
     "execution_count": 1785,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.groupby(['groupe', 'quizz'])[\"note\"].mean().unstack()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Seems better ! "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1791,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>eleve</th>\n",
       "      <th>note</th>\n",
       "      <th>groupe</th>\n",
       "      <th>quizz</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [eleve, note, groupe, quizz]\n",
       "Index: []"
      ]
     },
     "execution_count": 1791,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes[(df_notes.groupe == 'ibo2') & (df_notes.quizz == 'td4')]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "seems we don't have data for the exam number 4 fro this group (which had been cancelled due too large manifestations in Paris which lead to postpone the session too late.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1940,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_notes = df_notes[~(df_notes.quizz=='td4')]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Pivot Table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1821,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>quizz</th>\n",
       "      <th>td1</th>\n",
       "      <th>td2</th>\n",
       "      <th>td3</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>groupe</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>ibo1</th>\n",
       "      <td>67.618000</td>\n",
       "      <td>94.666667</td>\n",
       "      <td>90.739630</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo2</th>\n",
       "      <td>76.846207</td>\n",
       "      <td>98.666667</td>\n",
       "      <td>95.678148</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo3</th>\n",
       "      <td>40.002000</td>\n",
       "      <td>88.666667</td>\n",
       "      <td>88.887037</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo4</th>\n",
       "      <td>56.632500</td>\n",
       "      <td>87.407407</td>\n",
       "      <td>87.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo5</th>\n",
       "      <td>53.967407</td>\n",
       "      <td>90.000000</td>\n",
       "      <td>90.475714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo6</th>\n",
       "      <td>61.427667</td>\n",
       "      <td>90.344828</td>\n",
       "      <td>83.332857</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo7</th>\n",
       "      <td>74.875517</td>\n",
       "      <td>85.714286</td>\n",
       "      <td>94.642500</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "quizz         td1        td2        td3\n",
       "groupe                                 \n",
       "ibo1    67.618000  94.666667  90.739630\n",
       "ibo2    76.846207  98.666667  95.678148\n",
       "ibo3    40.002000  88.666667  88.887037\n",
       "ibo4    56.632500  87.407407  87.500000\n",
       "ibo5    53.967407  90.000000  90.475714\n",
       "ibo6    61.427667  90.344828  83.332857\n",
       "ibo7    74.875517  85.714286  94.642500"
      ]
     },
     "execution_count": 1821,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.groupby(['groupe', 'quizz'])[\"note\"].mean().unstack()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1824,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>quizz</th>\n",
       "      <th>td1</th>\n",
       "      <th>td2</th>\n",
       "      <th>td3</th>\n",
       "      <th>All</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>groupe</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>ibo1</th>\n",
       "      <td>67.618000</td>\n",
       "      <td>94.666667</td>\n",
       "      <td>90.739630</td>\n",
       "      <td>84.120805</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo2</th>\n",
       "      <td>76.846207</td>\n",
       "      <td>98.666667</td>\n",
       "      <td>95.678148</td>\n",
       "      <td>90.370349</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo3</th>\n",
       "      <td>40.002000</td>\n",
       "      <td>88.666667</td>\n",
       "      <td>88.887037</td>\n",
       "      <td>84.838065</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo4</th>\n",
       "      <td>56.632500</td>\n",
       "      <td>87.407407</td>\n",
       "      <td>87.500000</td>\n",
       "      <td>77.056747</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo5</th>\n",
       "      <td>53.967407</td>\n",
       "      <td>90.000000</td>\n",
       "      <td>90.475714</td>\n",
       "      <td>78.711059</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo6</th>\n",
       "      <td>61.427667</td>\n",
       "      <td>90.344828</td>\n",
       "      <td>83.332857</td>\n",
       "      <td>78.116667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo7</th>\n",
       "      <td>74.875517</td>\n",
       "      <td>85.714286</td>\n",
       "      <td>94.642500</td>\n",
       "      <td>84.957412</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>All</th>\n",
       "      <td>64.686180</td>\n",
       "      <td>90.882353</td>\n",
       "      <td>90.154715</td>\n",
       "      <td>82.528696</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "quizz         td1        td2        td3        All\n",
       "groupe                                            \n",
       "ibo1    67.618000  94.666667  90.739630  84.120805\n",
       "ibo2    76.846207  98.666667  95.678148  90.370349\n",
       "ibo3    40.002000  88.666667  88.887037  84.838065\n",
       "ibo4    56.632500  87.407407  87.500000  77.056747\n",
       "ibo5    53.967407  90.000000  90.475714  78.711059\n",
       "ibo6    61.427667  90.344828  83.332857  78.116667\n",
       "ibo7    74.875517  85.714286  94.642500  84.957412\n",
       "All     64.686180  90.882353  90.154715  82.528696"
      ]
     },
     "execution_count": 1824,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes.pivot_table('note', index='groupe', columns='quizz', margins=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1941,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr:last-of-type th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"6\" halign=\"left\">note</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"3\" halign=\"left\">max</th>\n",
       "      <th colspan=\"3\" halign=\"left\">min</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>quizz</th>\n",
       "      <th>td1</th>\n",
       "      <th>td2</th>\n",
       "      <th>td3</th>\n",
       "      <th>td1</th>\n",
       "      <th>td2</th>\n",
       "      <th>td3</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>groupe</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>ibo1</th>\n",
       "      <td>100.00</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>28.57</td>\n",
       "      <td>80.0</td>\n",
       "      <td>50.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo2</th>\n",
       "      <td>100.00</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>42.86</td>\n",
       "      <td>80.0</td>\n",
       "      <td>83.33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo3</th>\n",
       "      <td>57.14</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>14.29</td>\n",
       "      <td>40.0</td>\n",
       "      <td>66.67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo4</th>\n",
       "      <td>100.00</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>40.0</td>\n",
       "      <td>66.67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo5</th>\n",
       "      <td>85.71</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>60.0</td>\n",
       "      <td>50.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo6</th>\n",
       "      <td>100.00</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>28.57</td>\n",
       "      <td>40.0</td>\n",
       "      <td>50.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ibo7</th>\n",
       "      <td>100.00</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>28.57</td>\n",
       "      <td>40.0</td>\n",
       "      <td>66.67</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          note                                  \n",
       "           max                  min             \n",
       "quizz      td1    td2    td3    td1   td2    td3\n",
       "groupe                                          \n",
       "ibo1    100.00  100.0  100.0  28.57  80.0  50.00\n",
       "ibo2    100.00  100.0  100.0  42.86  80.0  83.33\n",
       "ibo3     57.14  100.0  100.0  14.29  40.0  66.67\n",
       "ibo4    100.00  100.0  100.0   0.00  40.0  66.67\n",
       "ibo5     85.71  100.0  100.0   0.00  60.0  50.00\n",
       "ibo6    100.00  100.0  100.0  28.57  40.0  50.00\n",
       "ibo7    100.00  100.0  100.0  28.57  40.0  66.67"
      ]
     },
     "execution_count": 1941,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results = df_notes.pivot_table(index='groupe', columns='quizz', \n",
    "                     aggfunc={\"note\":['max', min]})\n",
    "results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1837,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1842,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1855,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.rcParams['figure.figsize'] = (20, 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1857,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x11d11eba8>"
      ]
     },
     "execution_count": 1857,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "results['note'].plot(kind='bar')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1948,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x11ddc1cc0>"
      ]
     },
     "execution_count": 1948,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_notes.groupby('groupe').appreciation.value_counts(dropna=False).sort_values().plot(kind=\"bar\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1994,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x11d1b4908>"
      ]
     },
     "execution_count": 1994,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "(df_notes\n",
    "     .groupby(['quizz','groupe'])\n",
    "     ['appreciation']\n",
    "     .value_counts().sort_index()\n",
    "     .plot(kind=\"bar\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1995,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x11ef2ff98>"
      ]
     },
     "execution_count": 1995,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "(df_notes\n",
    "     .groupby(['quizz','groupe'])\n",
    "     ['appreciation']\n",
    "     .value_counts()\n",
    "     .unstack().plot(kind='bar', stacked=True))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1996,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x11ee1d7f0>"
      ]
     },
     "execution_count": 1996,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "(df_notes\n",
    "     .pivot_table(index='quizz', \n",
    "                  columns='groupe', \n",
    "                  aggfunc={'appreciation':'value_counts'})\n",
    "     .plot(kind=\"bar\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1945,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x11d5a9128>"
      ]
     },
     "execution_count": 1945,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_notes.appreciation.value_counts(dropna=False).plot(kind=\"bar\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1946,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>eleve</th>\n",
       "      <th>note</th>\n",
       "      <th>groupe</th>\n",
       "      <th>quizz</th>\n",
       "      <th>appreciation</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>75</th>\n",
       "      <td>Eleve183</td>\n",
       "      <td>0.0</td>\n",
       "      <td>ibo4</td>\n",
       "      <td>td1</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>525</th>\n",
       "      <td>Eleve111</td>\n",
       "      <td>0.0</td>\n",
       "      <td>ibo5</td>\n",
       "      <td>td1</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        eleve  note groupe quizz appreciation\n",
       "75   Eleve183   0.0   ibo4   td1          NaN\n",
       "525  Eleve111   0.0   ibo5   td1          NaN"
      ]
     },
     "execution_count": 1946,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_notes[pd.isna(df_notes.appreciation)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "missing grades for students although in the table at practice session1, maybe due to possible not good scrapping ?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Final question should be...\n",
    "\n",
    "## What should we put in `fillna` for them 😉 ?"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}