Github_Assignment_ML_L6.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Github_Assignment_ML_L6.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "blXAIWH0h5QX",
        "colab_type": "text"
      },
      "source": [
        "# Assignment on Feature Engineering (L6)\n",
        "\n",
        "This assignment has been adapted from the course **Feature Engineering for Machine Learning in Python** On DataCamp. \n",
        "\n",
        "We will explore what feature engineering is and how to get started with applying it to real-world data. We will be working with a modified subset of the [Stackoverflow survey response data](https://insights.stackoverflow.com/survey/2018/#overview). This data set records the details, and preferences of thousands of users of the StackOverflow website. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "V8OJHHSFhkVo",
        "colab_type": "text"
      },
      "source": [
        "## Creating Features \n",
        "---"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xxrLsTxvi4Ef",
        "colab_type": "code",
        "outputId": "15c71f6b-5b4f-440c-b701-e99508761227",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "import pandas as pd \n",
        "import numpy as np \n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns "
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/statsmodels/tools/_testing.py:19: FutureWarning: pandas.util.testing is deprecated. Use the functions in the public API at pandas.testing instead.\n",
            "  import pandas.util.testing as tm\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lw55w5YsB6jE",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Inspect your data\n",
        "---\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fCpuZS41h35f",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Load the data which has been stored as a CSV on the URL given below:\n",
        "so_survey_csv = 'https://assets.datacamp.com/production/repositories/3752/datasets/19699a2441073ad6459bf5e3e17690e2cae86cf1/Combined_DS_v10.csv'\n",
        "\n",
        "# Import so_survey_csv into so_survey_df\n",
        "so_survey_df = pd.read_csv(so_survey_csv)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yO7TR7EGBgDP",
        "colab_type": "text"
      },
      "source": [
        "Instructions: For the `so_survey_df` DataFrame, \n",
        "* Print its shape and its first five rows. \n",
        "* Print the data type of each column.  "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "09wAgH8u3_Kl",
        "colab_type": "code",
        "outputId": "7cb49b6c-8c87-462b-ce10-60b7df24fe6d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 374
        }
      },
      "source": [
        "print('shape:', so_survey_df.shape)\n",
        "print('head:', so_survey_df.head())\n",
        "print('Data types', so_survey_df.dtypes)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "shape: (999, 11)\n",
            "head:       SurveyDate  ...   RawSalary\n",
            "0  2/28/18 20:20  ...         NaN\n",
            "1  6/28/18 13:26  ...   70,841.00\n",
            "2    6/6/18 3:37  ...         NaN\n",
            "3    5/9/18 1:06  ...   21,426.00\n",
            "4  4/12/18 22:41  ...  £41,671.00\n",
            "\n",
            "[5 rows x 11 columns]\n",
            "Data types SurveyDate                     object\n",
            "FormalEducation                object\n",
            "ConvertedSalary               float64\n",
            "Hobby                          object\n",
            "Country                        object\n",
            "StackOverflowJobsRecommend    float64\n",
            "VersionControl                 object\n",
            "Age                             int64\n",
            "Years Experience                int64\n",
            "Gender                         object\n",
            "RawSalary                      object\n",
            "dtype: object\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1MzXp3T8Cl8W",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Selecting specific data types\n",
        "---\n",
        "Often a data set will contain columns with several different data types (like the one we are working with). The majority of machine learning models require us to have a consistent data type across features. Similarly, most feature engineering techniques are applicable to only one type of data at a time. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "glnEQlrVCobL",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Create a subset of `so_survey_df` consisting of only the numeric (int and float) columns and save it as `so_numeric_df`. \n",
        "* Print the column names contained in `so_numeric_df`. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sB3pYXsc5J7c",
        "colab_type": "code",
        "outputId": "27e3da46-9767-492e-f45c-2c448c64d73c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "source": [
        "so_survey_df_1= so_survey_df.select_dtypes(include=('int', 'float'))\n",
        "print(so_survey_df_1.columns)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Index(['ConvertedSalary', 'StackOverflowJobsRecommend', 'Age',\n",
            "       'Years Experience'],\n",
            "      dtype='object')\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "O_4kpncQDXuj",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### One-hot encoding and dummy variables\n",
        "---\n",
        "To use categorical variables in a machine learning model, we first need to represent them in a quantitative way. The two most common approaches are to one-hot encode the variables using or to use dummy variables. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oRQM6bVaDhnT",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* One-hot encode the `Country` column of `so_survey_df` DataFrame, adding \"OH\" as a prefix for each column.\n",
        "* Create dummy variables for the `Country` column, adding \"DM\" as a prefix for each column."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "f2wjdtHP6sM5",
        "colab_type": "code",
        "outputId": "6fbca3a3-7486-4e7e-a527-af5f779de11b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 530
        }
      },
      "source": [
        "one_hot_encoded = pd.get_dummies(so_survey_df, columns=['Country'], prefix='OH')\n",
        "print(one_hot_encoded.columns)\n",
        "one_hot_encoded.head()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Index(['SurveyDate', 'FormalEducation', 'ConvertedSalary', 'Hobby',\n",
            "       'StackOverflowJobsRecommend', 'VersionControl', 'Age',\n",
            "       'Years Experience', 'Gender', 'RawSalary', 'OH_France', 'OH_India',\n",
            "       'OH_Ireland', 'OH_Russia', 'OH_South Africa', 'OH_Spain', 'OH_Sweeden',\n",
            "       'OH_UK', 'OH_USA', 'OH_Ukraine'],\n",
            "      dtype='object')\n"
          ],
          "name": "stdout"
        },
        {
          "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>SurveyDate</th>\n",
              "      <th>FormalEducation</th>\n",
              "      <th>ConvertedSalary</th>\n",
              "      <th>Hobby</th>\n",
              "      <th>StackOverflowJobsRecommend</th>\n",
              "      <th>VersionControl</th>\n",
              "      <th>Age</th>\n",
              "      <th>Years Experience</th>\n",
              "      <th>Gender</th>\n",
              "      <th>RawSalary</th>\n",
              "      <th>OH_France</th>\n",
              "      <th>OH_India</th>\n",
              "      <th>OH_Ireland</th>\n",
              "      <th>OH_Russia</th>\n",
              "      <th>OH_South Africa</th>\n",
              "      <th>OH_Spain</th>\n",
              "      <th>OH_Sweeden</th>\n",
              "      <th>OH_UK</th>\n",
              "      <th>OH_USA</th>\n",
              "      <th>OH_Ukraine</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2/28/18 20:20</td>\n",
              "      <td>Bachelor's degree (BA. BS. B.Eng.. etc.)</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Yes</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Git</td>\n",
              "      <td>21</td>\n",
              "      <td>13</td>\n",
              "      <td>Male</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>6/28/18 13:26</td>\n",
              "      <td>Bachelor's degree (BA. BS. B.Eng.. etc.)</td>\n",
              "      <td>70841.0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>7.0</td>\n",
              "      <td>Git;Subversion</td>\n",
              "      <td>38</td>\n",
              "      <td>9</td>\n",
              "      <td>Male</td>\n",
              "      <td>70,841.00</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>6/6/18 3:37</td>\n",
              "      <td>Bachelor's degree (BA. BS. B.Eng.. etc.)</td>\n",
              "      <td>NaN</td>\n",
              "      <td>No</td>\n",
              "      <td>8.0</td>\n",
              "      <td>Git</td>\n",
              "      <td>45</td>\n",
              "      <td>11</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>5/9/18 1:06</td>\n",
              "      <td>Some college/university study without earning ...</td>\n",
              "      <td>21426.0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Zip file back-ups</td>\n",
              "      <td>46</td>\n",
              "      <td>12</td>\n",
              "      <td>Male</td>\n",
              "      <td>21,426.00</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>4/12/18 22:41</td>\n",
              "      <td>Bachelor's degree (BA. BS. B.Eng.. etc.)</td>\n",
              "      <td>41671.0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>8.0</td>\n",
              "      <td>Git</td>\n",
              "      <td>39</td>\n",
              "      <td>7</td>\n",
              "      <td>Male</td>\n",
              "      <td>£41,671.00</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      SurveyDate  ... OH_Ukraine\n",
              "0  2/28/18 20:20  ...          0\n",
              "1  6/28/18 13:26  ...          0\n",
              "2    6/6/18 3:37  ...          0\n",
              "3    5/9/18 1:06  ...          0\n",
              "4  4/12/18 22:41  ...          0\n",
              "\n",
              "[5 rows x 20 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eLKa7pey6vGv",
        "colab_type": "code",
        "outputId": "c6660fd6-31eb-4daa-af1c-d29ef6c39eaa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 530
        }
      },
      "source": [
        "dummy=pd.get_dummies(so_survey_df, columns=['Country'], prefix='DM')\n",
        "print(dummy.columns)\n",
        "dummy.head()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Index(['SurveyDate', 'FormalEducation', 'ConvertedSalary', 'Hobby',\n",
            "       'StackOverflowJobsRecommend', 'VersionControl', 'Age',\n",
            "       'Years Experience', 'Gender', 'RawSalary', 'DM_France', 'DM_India',\n",
            "       'DM_Ireland', 'DM_Russia', 'DM_South Africa', 'DM_Spain', 'DM_Sweeden',\n",
            "       'DM_UK', 'DM_USA', 'DM_Ukraine'],\n",
            "      dtype='object')\n"
          ],
          "name": "stdout"
        },
        {
          "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>SurveyDate</th>\n",
              "      <th>FormalEducation</th>\n",
              "      <th>ConvertedSalary</th>\n",
              "      <th>Hobby</th>\n",
              "      <th>StackOverflowJobsRecommend</th>\n",
              "      <th>VersionControl</th>\n",
              "      <th>Age</th>\n",
              "      <th>Years Experience</th>\n",
              "      <th>Gender</th>\n",
              "      <th>RawSalary</th>\n",
              "      <th>DM_France</th>\n",
              "      <th>DM_India</th>\n",
              "      <th>DM_Ireland</th>\n",
              "      <th>DM_Russia</th>\n",
              "      <th>DM_South Africa</th>\n",
              "      <th>DM_Spain</th>\n",
              "      <th>DM_Sweeden</th>\n",
              "      <th>DM_UK</th>\n",
              "      <th>DM_USA</th>\n",
              "      <th>DM_Ukraine</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2/28/18 20:20</td>\n",
              "      <td>Bachelor's degree (BA. BS. B.Eng.. etc.)</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Yes</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Git</td>\n",
              "      <td>21</td>\n",
              "      <td>13</td>\n",
              "      <td>Male</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>6/28/18 13:26</td>\n",
              "      <td>Bachelor's degree (BA. BS. B.Eng.. etc.)</td>\n",
              "      <td>70841.0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>7.0</td>\n",
              "      <td>Git;Subversion</td>\n",
              "      <td>38</td>\n",
              "      <td>9</td>\n",
              "      <td>Male</td>\n",
              "      <td>70,841.00</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>6/6/18 3:37</td>\n",
              "      <td>Bachelor's degree (BA. BS. B.Eng.. etc.)</td>\n",
              "      <td>NaN</td>\n",
              "      <td>No</td>\n",
              "      <td>8.0</td>\n",
              "      <td>Git</td>\n",
              "      <td>45</td>\n",
              "      <td>11</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>5/9/18 1:06</td>\n",
              "      <td>Some college/university study without earning ...</td>\n",
              "      <td>21426.0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Zip file back-ups</td>\n",
              "      <td>46</td>\n",
              "      <td>12</td>\n",
              "      <td>Male</td>\n",
              "      <td>21,426.00</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>4/12/18 22:41</td>\n",
              "      <td>Bachelor's degree (BA. BS. B.Eng.. etc.)</td>\n",
              "      <td>41671.0</td>\n",
              "      <td>Yes</td>\n",
              "      <td>8.0</td>\n",
              "      <td>Git</td>\n",
              "      <td>39</td>\n",
              "      <td>7</td>\n",
              "      <td>Male</td>\n",
              "      <td>£41,671.00</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      SurveyDate  ... DM_Ukraine\n",
              "0  2/28/18 20:20  ...          0\n",
              "1  6/28/18 13:26  ...          0\n",
              "2    6/6/18 3:37  ...          0\n",
              "3    5/9/18 1:06  ...          0\n",
              "4  4/12/18 22:41  ...          0\n",
              "\n",
              "[5 rows x 20 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UZ56-6j9oz4H",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Dealing with uncommon categories\n",
        "---\n",
        "Some features can have many different categories but a very uneven distribution of their occurrences. Take for example Data Science's favorite languages to code in, some common choices are Python, R, and Julia, but there can be individuals with bespoke choices, like FORTRAN, C etc. In these cases, we may not want to create a feature for each value, but only the more common occurrences."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1z1TCKQ7EHcb",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Extract the `Country` column of `so_survey_df` as a series and assign it to `countries`.\n",
        "* Find the counts of each category in the newly created `countries` series."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h16N9V7VGkcn",
        "colab_type": "code",
        "outputId": "f8d5c7bf-6fcb-4985-c056-e17918d5f7eb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        }
      },
      "source": [
        "countries = so_survey_df.Country\n",
        "country_counts=countries.value_counts()\n",
        "print(country_counts)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "South Africa    166\n",
            "USA             164\n",
            "Spain           134\n",
            "Sweeden         119\n",
            "France          115\n",
            "Russia           97\n",
            "India            95\n",
            "UK               95\n",
            "Ukraine           9\n",
            "Ireland           5\n",
            "Name: Country, dtype: int64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lKxW1Z70EW5W",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Create a mask for values occurring less than 10 times in `country_counts`.\n",
        "* Print the first 5 rows of the mask."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hCgOUaiZ6vfY",
        "colab_type": "code",
        "outputId": "d5111fb2-5854-4ee7-c837-c85df7c35428",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "mask= countries.isin(country_counts[country_counts<10].index)\n",
        "print(mask.head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0    False\n",
            "1    False\n",
            "2    False\n",
            "3    False\n",
            "4    False\n",
            "Name: Country, dtype: bool\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6Zro-imOEiwt",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Label values occurring less than the `mask` cutoff as 'Other'.\n",
        "* Print the new category counts in `countries`."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dMaEeRwr6v_X",
        "colab_type": "code",
        "outputId": "fc49e99f-1c64-44ae-bcc1-9e533707481c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 272
        }
      },
      "source": [
        "countries[mask]= 'Other'\n",
        "print(countries.value_counts())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "South Africa    166\n",
            "USA             164\n",
            "Spain           134\n",
            "Sweeden         119\n",
            "France          115\n",
            "Russia           97\n",
            "India            95\n",
            "UK               95\n",
            "Other            14\n",
            "Name: Country, dtype: int64\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:1: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  \"\"\"Entry point for launching an IPython kernel.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "w0j7IlzLpke5",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Binarizing columns\n",
        "---\n",
        "While numeric values can often be used without any feature engineering, there will be cases when some form of manipulation can be useful. For example on some occasions, we might not care about the magnitude of a value but only care about its direction, or if it exists at all. In these situations, we will want to binarize a column. In the `so_survey_df` data, we have a large number of survey respondents that are working voluntarily (without pay). We will create a new column titled `Paid_Job` indicating whether each person is paid (their salary is greater than zero)."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HdCPaTzFExzS",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Create a new column called `Paid_Job` filled with zeros.\n",
        "* Replace all the `Paid_Job` values with a 1 where the corresponding `ConvertedSalary` is greater than 0."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zHKF5uUD6wm_",
        "colab_type": "code",
        "outputId": "c3566255-bce6-4432-d6c0-ad7a306b15b7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "so_survey_df['Paid_Job'] = 0\n",
        "so_survey_df.loc[so_survey_df['ConvertedSalary']>0, 'Paid_Job'] = 1\n",
        "print(so_survey_df[['Paid_Job', 'ConvertedSalary']].head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "   Paid_Job  ConvertedSalary\n",
            "0         0              NaN\n",
            "1         1          70841.0\n",
            "2         0              NaN\n",
            "3         1          21426.0\n",
            "4         1          41671.0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7lJTmpvzq_NJ",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Binning values\n",
        "---\n",
        "For many continuous values we will care less about the exact value of a numeric column, but instead care about the bucket it falls into. This can be useful when plotting values, or simplifying your machine learning models. It is mostly used on continuous variables where accuracy is not the biggest concern e.g. age, height, wages.\n",
        "\n",
        "Bins are created using `pd.cut(df['column_name'], bins)` where bins can be an integer specifying the number of evenly spaced bins, or a list of bin boundaries."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GglGS3NkFtU3",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Bin the value of the `ConvertedSalary` column in `so_survey_df` into 5 equal bins, in a new column called `equal_binned`.\n",
        "* Print the first five rows of both columns: `ConvertedSalary` and `equal_binned`. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GYBMjVHS6xyf",
        "colab_type": "code",
        "outputId": "48f12653-e7bf-4a86-d7df-27dcfd40922e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "so_survey_df['equal_binned'] = pd.cut(so_survey_df['ConvertedSalary'], 5)\n",
        "print(so_survey_df[['equal_binned', 'ConvertedSalary']].head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "          equal_binned  ConvertedSalary\n",
            "0                  NaN              NaN\n",
            "1  (-2000.0, 400000.0]          70841.0\n",
            "2                  NaN              NaN\n",
            "3  (-2000.0, 400000.0]          21426.0\n",
            "4  (-2000.0, 400000.0]          41671.0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-o_1_UiiF1dc",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Bin the `ConvertedSalary` column using the boundaries in the list bins and label the bins using `labels` in a new column called `boundary_binned`. \n",
        "* Print the first 5 rows of the `boundary_binned` column. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UZyNvV6krVLJ",
        "colab_type": "code",
        "outputId": "8b2622c7-1d4f-4979-d59d-1ac269e65f52",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "# Specify the boundaries of the bins\n",
        "bins = [-np.inf, 10000, 50000, 100000, 150000, np.inf]\n",
        "\n",
        "# Bin labels\n",
        "labels = ['Very low', 'Low', 'Medium', 'High', 'Very high']\n",
        "\n",
        "so_survey_df['boundary_binned'] = pd.cut(so_survey_df['ConvertedSalary'], \n",
        "                                         bins= bins, labels=labels)\n",
        "print(so_survey_df[['boundary_binned', 'ConvertedSalary']].head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "  boundary_binned  ConvertedSalary\n",
            "0             NaN              NaN\n",
            "1          Medium          70841.0\n",
            "2             NaN              NaN\n",
            "3             Low          21426.0\n",
            "4             Low          41671.0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cIkFDAVUGhcQ",
        "colab_type": "text"
      },
      "source": [
        "## Dealing with Messy Data\n",
        "---"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "m8o-Z4YEsYF9",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### How sparse is my data?\n",
        "---\n",
        "Most data sets contain missing values, often represented as NaN (Not a Number). If we are working with Pandas, we can easily check how many missing values exist in each column.\n",
        "\n",
        "Let's find out how many of the developers taking the survey chose to enter their age (found in the `Age` column of `so_survey_df`) and their gender (`Gender` column of `so_survey_df`)."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1n0JDxzMHMwD",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Subset the `so_survey_df` DataFrame to only include the `Age` and `Gender` columns.\n",
        "* Print the number of non-missing values in both columns."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EByEYUTR6zUo",
        "colab_type": "code",
        "outputId": "a1d4b55b-c7ba-4959-96a0-20afce888237",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "source": [
        "sub_df= so_survey_df[['Age', 'Gender']]\n",
        "print(sub_df.notnull().sum())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Age       999\n",
            "Gender    693\n",
            "dtype: int64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "STDb03R3swA4",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Finding the missing values\n",
        "---\n",
        "While having a summary of how much of your data is missing can be useful, often we will need to find the exact locations of these missing values. Using the same subset of the StackOverflow data from the last exercise (`sub_df`), we will show how a value can be flagged as missing."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "E5XbMPSrHhY4",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Print the first 10 entries of the `sub_df` DataFrame.\n",
        "* Print the locations of the missing values in the first 10 rows of this DataFrame.\n",
        "* Print the locations of the non-missing values in the first 10 rows."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PAD53XAm6z6W",
        "colab_type": "code",
        "outputId": "fa24dd95-6087-45ce-c0ff-32c59334dace",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        }
      },
      "source": [
        "print(sub_df.head(10))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "   Age  Gender\n",
            "0   21    Male\n",
            "1   38    Male\n",
            "2   45     NaN\n",
            "3   46    Male\n",
            "4   39    Male\n",
            "5   39    Male\n",
            "6   34    Male\n",
            "7   24  Female\n",
            "8   23    Male\n",
            "9   36     NaN\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SlA7KQI7Avax",
        "colab_type": "code",
        "outputId": "57afad58-c55c-4a29-8246-2f9737922ad2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        }
      },
      "source": [
        "print(sub_df.head(10).isna())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "     Age  Gender\n",
            "0  False   False\n",
            "1  False   False\n",
            "2  False    True\n",
            "3  False   False\n",
            "4  False   False\n",
            "5  False   False\n",
            "6  False   False\n",
            "7  False   False\n",
            "8  False   False\n",
            "9  False    True\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Hf4Kga2fAfV_",
        "colab_type": "code",
        "outputId": "5f09ab4e-5bff-4b64-9d39-812d28adeb62",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        }
      },
      "source": [
        "print(sub_df.head(10).notnull())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "    Age  Gender\n",
            "0  True    True\n",
            "1  True    True\n",
            "2  True   False\n",
            "3  True    True\n",
            "4  True    True\n",
            "5  True    True\n",
            "6  True    True\n",
            "7  True    True\n",
            "8  True    True\n",
            "9  True   False\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CgR_uodEtS2o",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Listwise deletion\n",
        "---\n",
        "The simplest way to deal with missing values in our dataset when they are occurring entirely at random is to remove those rows, also called 'listwise deletion'.\n",
        "\n",
        "Depending on the use case, we will sometimes want to remove all missing values in our data while other times we may want to only remove a particular column if too many values are missing in that column."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kIioVg70H8fY",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Print the number of rows and columns in `so_survey_df`.\n",
        "* Drop all rows with missing values in `so_survey_df`.\n",
        "* Drop all columns with missing values in `so_survey_df`.\n",
        "* Drop all rows in `so_survey_df` where `Gender` is missing."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l1yyAGA6603W",
        "colab_type": "code",
        "outputId": "81450045-50ef-493a-f7f6-64ec3d56c66e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "print(so_survey_df.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(999, 14)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mH8wqgq4BJxj",
        "colab_type": "code",
        "outputId": "47725d25-661b-4e6e-ee3c-add5f73a78a3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "so_survey_df_2= so_survey_df.dropna()\n",
        "print(so_survey_df_2.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(264, 14)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FaL9d-P8BtYy",
        "colab_type": "code",
        "outputId": "1b5f6508-2951-48e7-bf8d-6ca9dc66dd6c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "so_survey_df_3= so_survey_df.dropna(axis=1, how='any')\n",
        "print(so_survey_df_3.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(999, 8)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ThjooRRGCWPR",
        "colab_type": "code",
        "outputId": "587864bd-3619-467e-8b69-1a20421630cd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "no_gender= so_survey_df.dropna(subset=['Gender'])\n",
        "print(no_gender.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(693, 14)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9C1H8JwhtxHU",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Replacing missing values with constants\n",
        "---\n",
        "While removing missing data entirely maybe a correct approach in many situations, this may result in a lot of information being omitted from your models.\n",
        "\n",
        "We may find categorical columns where the missing value is a valid piece of information in itself, such as someone refusing to answer a question in a survey. In these cases, we can fill all missing values with a new category entirely, for example 'No response given'."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cPq_Ae40IUnY",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Print the count of occurrences of each category in `so_survey_df`'s `Gender` column.\n",
        "* Replace all missing values in the `Gender` column with the string 'Not Given'. Make changes to the original DataFrame.\n",
        "* Print the count of occurrences of updated category in `so_survey_df`'s `Gender` column."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "z_ZvEJig61du",
        "colab_type": "code",
        "outputId": "4137e5c2-0486-4f63-eaa1-6611a2bb2851",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 170
        }
      },
      "source": [
        "print(so_survey_df['Gender'].value_counts())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Male                                                                         632\n",
            "Female                                                                        53\n",
            "Female;Male                                                                    2\n",
            "Transgender                                                                    2\n",
            "Male;Non-binary. genderqueer. or gender non-conforming                         1\n",
            "Female;Transgender                                                             1\n",
            "Non-binary. genderqueer. or gender non-conforming                              1\n",
            "Female;Male;Transgender;Non-binary. genderqueer. or gender non-conforming      1\n",
            "Name: Gender, dtype: int64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cqvE8m0GC55_",
        "colab_type": "code",
        "outputId": "a69f8cb0-1bd6-4e15-e8c7-55cceea5b341",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        }
      },
      "source": [
        "so_survey_df['Gender'].fillna('Not Given', inplace=True)\n",
        "print(so_survey_df['Gender'].value_counts())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Male                                                                         632\n",
            "Not Given                                                                    306\n",
            "Female                                                                        53\n",
            "Female;Male                                                                    2\n",
            "Transgender                                                                    2\n",
            "Male;Non-binary. genderqueer. or gender non-conforming                         1\n",
            "Female;Transgender                                                             1\n",
            "Non-binary. genderqueer. or gender non-conforming                              1\n",
            "Female;Male;Transgender;Non-binary. genderqueer. or gender non-conforming      1\n",
            "Name: Gender, dtype: int64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Rn2h-F8IuI0C",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Filling continuous missing values\n",
        "---\n",
        "Earlier, we dealt with different methods of removing data missing values and filling in missing values with a fixed string. These approaches are valid in many cases, particularly when dealing with categorical columns but have limited use when working with continuous values. In these cases, it may be most valid to fill the missing values in the column with a value calculated from the entries present in the column."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_S9YZ2GsIyLx",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Print the first five rows of the `StackOverflowJobsRecommend` column of `so_survey_df`.\n",
        "* Replace the missing values in the `StackOverflowJobsRecommend` column with its mean. Make changes directly to the original DataFrame.\n",
        "* Round the decimal values that we introduced in the `StackOverflowJobsRecommend` column."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NJ4dP6-i62LO",
        "colab_type": "code",
        "outputId": "bb8e632a-b805-44d3-80cb-387401dbf506",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "print(so_survey_df.StackOverflowJobsRecommend.head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0    NaN\n",
            "1    7.0\n",
            "2    8.0\n",
            "3    NaN\n",
            "4    8.0\n",
            "Name: StackOverflowJobsRecommend, dtype: float64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pDb7G_rSPjoc",
        "colab_type": "code",
        "outputId": "c84f3888-1b26-44a5-b74c-d8071d531659",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "so_survey_df['StackOverflowJobsRecommend'].fillna(so_survey_df['StackOverflowJobsRecommend'].mean(), inplace=True)\n",
        "print(so_survey_df['StackOverflowJobsRecommend'].head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0    7.061602\n",
            "1    7.000000\n",
            "2    8.000000\n",
            "3    7.061602\n",
            "4    8.000000\n",
            "Name: StackOverflowJobsRecommend, dtype: float64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SWT8VyA_QOs1",
        "colab_type": "code",
        "outputId": "5dca44c0-f528-40ed-a1c7-2a43ec5a6ec6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "so_survey_df['StackOverflowJobsRecommend']= round(so_survey_df['StackOverflowJobsRecommend'])\n",
        "print(so_survey_df['StackOverflowJobsRecommend'].head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0    7.0\n",
            "1    7.0\n",
            "2    8.0\n",
            "3    7.0\n",
            "4    8.0\n",
            "Name: StackOverflowJobsRecommend, dtype: float64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SKcC12f0vCuh",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Dealing with stray characters (I)\n",
        "---\n",
        "In this exercise, we will work with the `RawSalary` column of so_survey_df which contains the wages of the respondents along with the currency symbols and commas, such as $42,000. When importing data from Microsoft Excel, more often that not we will come across data in this form."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tz0CGSQkJI03",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Remove the commas (,) from the `RawSalary` column.\n",
        "* Remove the dollar ($) signs from the `RawSalary` column.\n",
        "* Print the first five rows of updated `RawSalary` column. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gacu0KrG62p2",
        "colab_type": "code",
        "outputId": "217b7ac2-d94b-4145-94c2-c3242d4657e5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "so_survey_df['RawSalary']= so_survey_df['RawSalary'].str.replace(',','')\n",
        "so_survey_df['RawSalary']= so_survey_df['RawSalary'].str.replace('$','')\n",
        "so_survey_df['RawSalary']= so_survey_df['RawSalary'].str.replace('£','')\n",
        "print(so_survey_df['RawSalary'].head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0         NaN\n",
            "1    70841.00\n",
            "2         NaN\n",
            "3    21426.00\n",
            "4    41671.00\n",
            "Name: RawSalary, dtype: object\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UpayqY5IwMBl",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Dealing with stray characters (II)\n",
        "---\n",
        "In the last exercise, we could tell quickly based off of the `df.head()` call which characters were causing an issue. In many cases this will not be so apparent. There will often be values deep within a column that are preventing us from casting a column as a numeric type so that it can be used in a model or further feature engineering.\n",
        "\n",
        "One approach to finding these values is to force the column to the data type desired using `pd.to_numeric()`, coercing any values causing issues to `NaN`, Then filtering the DataFrame by just the rows containing the `NaN` values.\n",
        "\n",
        "Try to cast the `RawSalary` column as a float and it will fail as an additional character can now be found in it. Find the character and remove it so the column can be cast as a float."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Z7Duf4v7JpPv",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Attempt to convert the `RawSalary` column of `so_survey_df` to numeric values coercing all failures into null values.\n",
        "* Find the indexes of the rows containing `NaN`s.\n",
        "* Print the rows in `RawSalary` based on these indexes."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AYuh53vbv5_d",
        "colab_type": "code",
        "outputId": "553ad99c-62fe-428c-b4f2-9ce5a94998aa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 221
        }
      },
      "source": [
        "# Attempt to convert the column to numeric values\n",
        "numeric_vals = pd.to_numeric(so_survey_df['RawSalary'], errors='coerce')\n",
        "idx= numeric_vals.isna()\n",
        "print(so_survey_df['RawSalary'][idx])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0      NaN\n",
            "2      NaN\n",
            "6      NaN\n",
            "8      NaN\n",
            "11     NaN\n",
            "      ... \n",
            "989    NaN\n",
            "990    NaN\n",
            "992    NaN\n",
            "994    NaN\n",
            "997    NaN\n",
            "Name: RawSalary, Length: 334, dtype: object\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "feXhsa94wtBz",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Method chaining\n",
        "---\n",
        "When applying multiple operations on the same column (like in the previous exercises), you made the changes in several steps, assigning the results back in each step. However, when applying multiple successive operations on the same column, you can \"chain\" these operations together for clarity and ease of management. This can be achieved by calling multiple methods sequentially:\n",
        "```\n",
        "# Method chaining\n",
        "df['column'] = df['column'].method1().method2().method3()\n",
        "\n",
        "# Same as \n",
        "df['column'] = df['column'].method1()\n",
        "df['column'] = df['column'].method2()\n",
        "df['column'] = df['column'].method3()\n",
        "```"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yo-JUG77KTRi",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Remove the commas (`,`) from the `RawSalary` column of `so_survey_df`.\n",
        "* Remove the dollar (`$`) signs from the `RawSalary` column.\n",
        "* Remove the pound (`£`) signs from the `RawSalary` column.\n",
        "* Convert the `RawSalary` column to float."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1XG2AjsJSZNt",
        "colab_type": "code",
        "outputId": "da434493-7792-4965-94e5-96a630b5981d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "so_survey_df['RawSalary']= so_survey_df['RawSalary']\\\n",
        ".str.replace(',','')\\\n",
        ".str.replace('$','')\\\n",
        ".str.replace('£','')\\\n",
        ".astype(float)\n",
        "print(so_survey_df['RawSalary'].head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0        NaN\n",
            "1    70841.0\n",
            "2        NaN\n",
            "3    21426.0\n",
            "4    41671.0\n",
            "Name: RawSalary, dtype: float64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CqpCNv72Kl_1",
        "colab_type": "text"
      },
      "source": [
        "## Conforming to Statistical Assumptions\n",
        "---"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hYwlN2FyyBml",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### What does your data look like?\n",
        "---\n",
        "Up until now we have focused on creating new features and dealing with issues in our data. Feature engineering can also be used to make the most out of the data that we already have and use it more effectively when creating machine learning models.\n",
        "Many algorithms may assume that our data is normally distributed, or at least that all our columns are on the same scale. This will often not be the case, e.g. one feature may be measured in thousands of dollars while another would be number of years. In this exercise, we will create plots to examine the distributions of some numeric columns in the `so_survey_df` DataFrame. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LHMvrvwxyTQu",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "so_numeric_df = so_survey_df[['ConvertedSalary', 'Age', 'Years Experience']]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "i8Ch5slQy6B-",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Generate a histogram of all columns in the `so_numeric_df` DataFrame.\n",
        "* Generate box plots of the `Age` and `Years Experience` columns in the `so_numeric_df` DataFrame.\n",
        "* Generate a box plot of the `ConvertedSalary` column in the `so_numeric_df`.\n",
        "* Plot pairwise relationships (using `sns.pairplot`) in the `so_numeric_df`."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g4HCQAhnTx0K",
        "colab_type": "code",
        "outputId": "12600884-46e6-4f6b-8dde-1a552dd7cd19",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 608
        }
      },
      "source": [
        "so_numeric_df.hist(figsize=(10,10))\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAlYAAAJPCAYAAAC+fJpMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAgAElEQVR4nOzde7ydZX3n/c9XgkrBcii6mwJD7JRqqRmRpohjDxt5VISOoR1lcBgBxUlnilan6bTRmVe1rc5D5yn1VLVPKlasCFKUQoVaGXTXw1NQUOSoNdpQEgNRxEiktUZ/zx/rii5isvdK9rX2Ifm8X6/12vd93affvdfKyndf973WlapCkiRJs/eI+S5AkiRpb2GwkiRJ6sRgJUmS1InBSpIkqRODlSRJUicGK0mSpE4MVpIkzbEkU0le0mE/5yb5eI+a1IfBSl20N4kHkjxqvmuRtLgl+Y9JbkqyNcmmJH+d5Ofmu67tkixLUkmWdNrfI5NcmGRDO+f1Sd7QY9+aewYrzVqSZcDPAwU8d16LkbSoJfkN4A3A/wImgH8FvBVYOZ91bdcrTO3glcAK4ATgMcAk8OkxHOdhxnQu+zyDlXo4G7gBeCdwzvbGJD+S5K+SfCPJp5K8drjLOskTk1yX5GtJPp/kjLkvXdJCkeRg4PeA86vq/VX1zar6dlX9VVX99ySPSvKGJF9ujzds7yVPMtl6fFYn2dx6ul7Ulj01yb1J9hs61i8nubVNPyLJmiRfTHJ/ksuTHNaWbe+dOi/JPwIfBj7advP11sP0tLbui5Pc1Xrv/ybJ0UPHe2aSzyXZkuSPgQyd+s8CV1bVl2tgfVW9a2jb7bU9mOTOJL88ze/wjUnuae+7Nyf5+aFlr0lyRZJ3J/kGsCbJQ0l+ZGid45N8Jcn+u/n0qTFYqYezgUva49lJJlr7W4BvAj/KIHANh64DgeuA9wCPA84E3prk2DmsW9LC8jTg0cCVu1j+P4ATgeOAJzPo4fmfQ8t/FDgYOAI4D3hLkkOr6kYG70XPGFr3PzJ4/wF4GXA68IvAjwEPMHj/GvaLwE8BzwZ+obUdUlUHVdXfJVkJvAr4FeCxwMeASwGSHA68v9V6OPBF4OlD+74B+I0kv5ZkeZLh0EVb/+fbuf0u8O4kS3fxO/pU+/0c1s7vL5I8emj5SuAK4BDgQmAKGP6j9oXAZVX17V3sXzOpKh8+9vgB/BzwbeDwNv854L8B+7X2Jwyt+1rg4236PwAf22Ff/y/w6vk+Jx8+fMzPAzgLuHea5V8ETh2afzawvk1PAv8ELBlavhk4sU2/FnhHm34Mg6B1dJu/Czh5aLul7f1rCbCMwW0OPz60fHvb8LH+GjhvaP4RwEPA0bRe/aFlATYAL2nz+wHnA58AvgV8GThnmt/DLcDKNn3u9vfVXaz7APDkNv0a4KM7LP8PwCeG6rgXOGG+XwuL+WGPlWbrHOBDVfXVNv+e1vZYBm9K9wytOzx9NPDUJF/f/mDwpvqjc1CzpIXpfuDwae79+THg7qH5u1vb97avqm1D8w8BB7Xp9wC/0i4d/grw6aravq+jgSuH3ovuAr7D4B6v7Ybfv3bmaOCNQ/v4GoMAdUSr8Xvb1yDFDM9/p6reUlVPZ9CT9DrgHUl+CiDJ2UluGdr3kxj0fP2AJL/ZLkduaesevMO6O57HVcCxSR4PPBPYUlWfnOFcNQ1vXNMeS3IAgy7k/ZLc25ofxeCNYQLYBhwJ/H1bdtTQ5vcAf1tVz5yjciUtfH/HoMfmdAaXq3b0ZQYB5o42/69a24yq6s4kdwPP4eGXAWHwfvTiqvrEjtu1D+fAoIeKnUwP7+N1VXXJTvZxDEPvf+1S31E7rtfq/CcGlzB/l0HgeQj4U+Bk4O+q6jtJbuHh92ht3+/PA7/V1r2jqr6b5IEd1n1Y7VX1z0kuB/4T8ETgz3dWl0Znj5Vm43QGf9Udy+Ca/nEM7kH4GIOu7/cDr0nyQ0me2Nq2+wDwk0lemGT/9vjZ7X+hSdr3VNUW4HcYBIvT23vH/kmek+R/M7hn6X8meWy7b+l3gHfvxiHeA7ycwT1SfzHU/ifA67bfbN72P92nEL8CfBf48R328cokP932cXCS57dl1wA/neRXWm/crzPUO5/kFe3m+wOSLElyDoPLlZ8BDmQQhr7S1n0Rgx6rnXkMgz9ovwIsSfI7wA9P+xsZeBeDS4rPxWA1awYrzcY5wJ9V1T9W1b3bH8AfM7is91IG3dD3MvjHeimDv0apqgeBZzG4af3LbZ0/YNDjJWkfVVUXAr/B4EbvrzDoCXop8JcM7pO6CbgVuI3BVxK8djd2fymDm9A/PHT7AsAbgauBDyV5kMHN5E+dpsaHGFyu+0S7PHdiVV3J4D3ssvaJu9sZ9I7RjvV84AIGlzuPYXA/1XYPMbiR/F7gqwzut/r3VfWlqrqzLfs74D5g+Q7bDvsb4IMMrhLcDfwzM1/CpPXUfZeHXx7VHkq7YU0auyR/APxoVZ0z48qSpDmT5MPAe6rq7fNdy2Jnj5XGJoPvqfo3GTiBwcefd/UxaknSPEjys8DxwHvnu5a9gTeva5wew6Dr/ccYdGFfyOATKJKkBSDJxQzul315u0VDszTjpcAkRzG4sW2CwQ10a6vqjUleA/xn2g11wKuq6tq2zSsZ9E58B/j1qvqb8ZQvSZK0cIwSrJYCS6vq00keA9zMIN2eAWytqj/cYf1jGfRSnMCgp+L/AD9ZVd8ZQ/2SJEkLxoz3WFXVpqr6dJt+kMEXpx0xzSYrGXwd/req6h+AdQxCliRJ0l5tt+6xal+U9hTgRgbjHL00ydkMPv66uqoeYBC6bhjabAPTBzEOP/zwWrZs2e6UslPf/OY3OfDAA2e9H+uwDuvYMzfffPNXq+qxXXe6l9md97uF8loZB89tcfLcBqZ7rxs5WCU5CHgf8Iqq+kaStwG/z+C+q99ncGPyi3djf6uAVQATExP84R/+4QxbzGzr1q0cdNBBM684ZtZhHftqHSeddJLfgTODZcuWcdNNN4207tTUFJOTk+MtaJ54bouT5zbQvsV/p0YKVkn2ZxCqLqmq9wNU1X1Dy/+UwTdpA2zk4V/Vf2Rre5iqWgusBVixYkX1eKIWyhNuHdZhHZK0b5rxHqs2ptFFwF1V9UdD7UuHVvtlBt8yC4Nvrz0zyaPaoI7HAA7oKEmS9nqj9Fg9HXghcFsb+BHgVcALkhzH4FLgeuBXAarqjjag450Mxiw6308ESpKkfcGMwaqqPs5ORtEGrp1mm9cxGEdJkiRpn+GQNpIkSZ0YrCRJkjoxWEmSJHVisJIkSerEYCVJktTJbg1psxAsW3PNLpetXr6Nc6dZPp31F5y2pyVJ0ljctnHLHr+nTcf3O2l87LGSJEnqxGAlSZLUicFKkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdWKwkqQmySFJrkjyuSR3JXlaksOSXJfkC+3noW3dJHlTknVJbk1y/HzXL2n+Gawk6fveCHywqp4IPBm4C1gDXF9VxwDXt3mA5wDHtMcq4G1zX66khcZgJUlAkoOBXwAuAqiqf6mqrwMrgYvbahcDp7fplcC7auAG4JAkS+e4bEkLjMFKkgYeD3wF+LMkn0ny9iQHAhNVtamtcy8w0aaPAO4Z2n5Da5O0D1t0gzBL0pgsAY4HXlZVNyZ5I9+/7AdAVVWS2t0dJ1nF4HIhExMTTE1NjbTdxAGDweV7G/X447R169YFUcc4eG6LU69zM1hJ0sAGYENV3djmr2AQrO5LsrSqNrVLfZvb8o3AUUPbH9nafkBVrQXWAqxYsaImJydHKujNl1zFhbf1f5tef9Zoxx+nqakpRv09LDae2+LU69y8FChJQFXdC9yT5Amt6WTgTuBq4JzWdg5wVZu+Gji7fTrwRGDL0CVDSfsoe6wk6fteBlyS5JHAl4AXMfgD9PIk5wF3A2e0da8FTgXWAQ+1dSXt4wxWktRU1S3Aip0sOnkn6xZw/tiLkrSoeClQkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdWKwkiRJ6sRgJUmS1InBSpIkqRODlSRJUicGK0mSpE4MVpIkSZ0YrCRJkjoxWEmSJHVisJIkSepkxmCV5KgkH0lyZ5I7kry8tR+W5LokX2g/D23tSfKmJOuS3Jrk+HGfhCRJ0kIwSo/VNmB1VR0LnAicn+RYYA1wfVUdA1zf5gGeAxzTHquAt3WvWpIkaQGaMVhV1aaq+nSbfhC4CzgCWAlc3Fa7GDi9Ta8E3lUDNwCHJFnavXJJkqQFZrfusUqyDHgKcCMwUVWb2qJ7gYk2fQRwz9BmG1qbJEnSXm3JqCsmOQh4H/CKqvpGku8tq6pKUrtz4CSrGFwqZGJigqmpqZG2W7182y6XTRww/fLpjHr8UWzdurXr/qzDOvbmOiRpbzJSsEqyP4NQdUlVvb8135dkaVVtapf6Nrf2jcBRQ5sf2doepqrWAmsBVqxYUZOTkyMVfO6aa3a5bPXybVx428hZ8WHWnzXa8UcxNTXFqOczTtZhHYuhDknam4zyqcAAFwF3VdUfDS26GjinTZ8DXDXUfnb7dOCJwJahS4aSJEl7rVG6d54OvBC4Lcktre1VwAXA5UnOA+4GzmjLrgVOBdYBDwEv6lqxJEnSAjVjsKqqjwPZxeKTd7J+AefPsi5JkqRFx29elyRJ6sRgJUmS1InBSpIkqRODlSQNSbI+yW1JbklyU2tzbFRJIzFYSdIPOqmqjquqFW3esVEljcRgJUkzc2xUSSMxWEnSwxXwoSQ3t6G3wLFRJY1oz8Z/kaS9189V1cYkjwOuS/K54YVzOTbqbMY/nc5CGCNybx6r0nNbnHqdm8FKkoZU1cb2c3OSK4ETmKexUd98yVV7PP7pdHqOjbqn9uaxKj23xanXuRmsxmzZNINGz8b6C04by36lfVmSA4FHVNWDbfpZwO/x/bFRL+AHx0Z9aZLLgKfi2KjSPs9gJUnfNwFcORh7niXAe6rqg0k+hWOjShqBwUqSmqr6EvDknbTfj2OjShqBnwqUJEnqxGAlSZLUicFKkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdWKwkiRJ6sRgJUmS1InBSpIkqRODlSRJUicGK0mSpE4MVpIkSZ0YrCRJkjoxWEmSJHVisJIkSerEYCVJktSJwUqSJKkTg5UkSVInBitJkqRODFaSJEmdGKwkSZI6MVhJkiR1YrCSJEnqxGAlSZLUyYzBKsk7kmxOcvtQ22uSbExyS3ucOrTslUnWJfl8kmePq3BJGock+yX5TJIPtPnHJ7mxva+9N8kjW/uj2vy6tnzZfNYtaWEYpcfqncApO2l/fVUd1x7XAiQ5FjgT+Om2zVuT7NerWEmaAy8H7hqa/wMG73c/ATwAnNfazwMeaO2vb+tJ2sfNGKyq6qPA10bc30rgsqr6VlX9A7AOOGEW9UnSnElyJHAa8PY2H+AZwBVtlYuB09v0yjZPW35yW1/SPmw291i9NMmt7VLhoa3tCOCeoXU2tDZJWgzeAPwW8N02/yPA16tqW5sffk/73vtdW76lrS9pH5aqmnmlwb0DH6iqJ7X5CeCrQAG/Dyytqhcn+WPghqp6d1vvIuCvq+qKnexzFbAKYGJi4mcuu+yykQq+beOWXS6bOADu+6eRdvMDlh9x8J5tuBNbt27loIMOAqavdzZGqXe4jvlkHftOHSeddNLNVbWi607nSJJfAk6tql9LMgn8JnAug/e0n2jrHMXgPe1J7b7TU6pqQ1v2ReCpVfXVnex7j97vNn9tyx6/p02n5/vdnloo/w7GwXNbnHbn3KZ7r1uyJwevqvu2Tyf5U+ADbXYjcNTQqke2tp3tYy2wFmDFihU1OTk50rHPXXPNLpetXr6NC2/bo1Ni/VmjHX8UU1NTbD+f6eqdjVHqHa5jPlmHdSwSTwee2z6M82jgh4E3AockWdJ6pYbf07a/321IsgQ4GLh/Zzve0/e7N19y1R6/p02n5/vdntqbX3+e2+LU69z26FJgkqVDs78MbP/E4NXAme3TMo8HjgE+ObsSJWn8quqVVXVkVS1j8CGcD1fVWcBHgOe11c4BrmrTV7d52vIP1yiXACTt1Wb8UyjJpcAkcHiSDcCrgckkxzG4FLge+FWAqrojyeXAncA24Pyq+s54SpekOfHbwGVJXgt8BriotV8E/HmSdQw+4HPmPNUnaQGZMVhV1Qt20nzRTtq2r/864HWzKUrzZ9kYLl2uXr6Nc9dcw/oLTuu+b2kcqmoKmGrTX2Inn26uqn8Gnj+nhUla8PzmdUmSpE4MVpIkSZ30/7iJ5sQol+y2X4KTJElzwx4rSZKkTgxWkiRJnXgpsOn5aTgvwUmStG+yx0qSJKkTg5UkSVInBitJkqRODFaSJEmdGKwkSZI6MVhJkiR1YrCSJEnqxGAlSZLUicFKkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdWKwkiRJ6sRgJUmS1InBSpIkqRODlSRJUicGK0mSpE4MVpIkSZ0YrCSpSfLoJJ9M8tkkdyT53db++CQ3JlmX5L1JHtnaH9Xm17Xly+azfknzz2AlSd/3LeAZVfVk4DjglCQnAn8AvL6qfgJ4ADivrX8e8EBrf31bT9I+zGAlSU0NbG2z+7dHAc8ArmjtFwOnt+mVbZ62/OQkmaNyJS1ABitJGpJkvyS3AJuB64AvAl+vqm1tlQ3AEW36COAegLZ8C/Ajc1uxpIVkyXwXIEkLSVV9BzguySHAlcATZ7vPJKuAVQATExNMTU2NtN3EAbB6+baZV9xNox5/nLZu3bog6hgHz21x6nVuBitJ2omq+nqSjwBPAw5JsqT1Sh0JbGyrbQSOAjYkWQIcDNy/k32tBdYCrFixoiYnJ0eq4c2XXMWFt/V/m15/1mjHH6epqSlG/T0sNp7b4tTr3LwUKElNkse2niqSHAA8E7gL+AjwvLbaOcBVbfrqNk9b/uGqqrmrWNJCY4+VJH3fUuDiJPsx+MPz8qr6QJI7gcuSvBb4DHBRW/8i4M+TrAO+Bpw5H0VLWjgMVpLUVNWtwFN20v4l4ISdtP8z8Pw5KE3SIuGlQEmSpE4MVpIkSZ0YrCRJkjqZMVgleUeSzUluH2o7LMl1Sb7Qfh7a2pPkTW3crFuTHD/O4iVJkhaSUXqs3gmcskPbGuD6qjoGuL7NAzwHOKY9VgFv61OmJEnSwjdjsKqqjzL4GPGw4fGxdhw3611tvK0bGHyp3tJexUqSJC1ke3qP1URVbWrT9wITbfp742Y1w2NqSZIk7dVm/T1WVVVJdvubhvd07Kzpxs0a17hau8s6dl7HfI8vtVDGuLIOSdp77Wmwui/J0qra1C71bW7t28fN2m54TK2H2dOxs85dc80ul61evm0s42rtLuvYeR3zPT7ZQhnjyjokae+1p5cCh8fH2nHcrLPbpwNPBLYMXTKUJEnaq83YnZHkUmASODzJBuDVwAXA5UnOA+4GzmirXwucCqwDHgJeNIaaJUmSFqQZg1VVvWAXi07eyboFnD/boiRJkhYjv3ldkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdWKwkiRJ6sRgJUmS1InBSpIkqRODlSRJUicGK0mSpE4MVpIkSZ0YrCRJkjoxWEmSJHVisJIkSerEYCVJQJKjknwkyZ1J7kjy8tZ+WJLrknyh/Ty0tSfJm5KsS3JrkuPn9wwkLQQGK0ka2AasrqpjgROB85McC6wBrq+qY4Dr2zzAc4Bj2mMV8La5L1nSQmOwkiSgqjZV1afb9IPAXcARwErg4rbaxcDpbXol8K4auAE4JMnSOS5b0gJjsJKkHSRZBjwFuBGYqKpNbdG9wESbPgK4Z2izDa1N0j5syXwXIEkLSZKDgPcBr6iqbyT53rKqqiS1B/tcxeByIRMTE0xNTY203cQBsHr5tt093IxGPf44bd26dUHUMQ6e2+LU69wMVpLUJNmfQai6pKre35rvS7K0qja1S32bW/tG4KihzY9sbT+gqtYCawFWrFhRk5OTI9Xz5kuu4sLb+r9Nrz9rtOOP09TUFKP+HhYbz21x6nVuBivNmWVrrhnLftdfcNpY9qt9SwZdUxcBd1XVHw0tuho4B7ig/bxqqP2lSS4DngpsGbpkKGkfZbCSpIGnAy8EbktyS2t7FYNAdXmS84C7gTPasmuBU4F1wEPAi+a2XEkLkcFKkoCq+jiQXSw+eSfrF3D+WIuStOj4qUBJkqRODFaSJEmdGKwkSZI6MVhJkiR1YrCSJEnqxGAlSZLUicFKkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdWKwkiRJ6sRgJUmS1MmS2WycZD3wIPAdYFtVrUhyGPBeYBmwHjijqh6YXZmSJEkLX48eq5Oq6riqWtHm1wDXV9UxwPVtXpIkaa83jkuBK4GL2/TFwOljOIYkSdKCM9tgVcCHktycZFVrm6iqTW36XmBilseQJElaFGZ1jxXwc1W1McnjgOuSfG54YVVVktrZhi2IrQKYmJhgampqpAOuXr5tl8smDph++VyxjrmtY9TXztatW0ded5ysQ5L2XrMKVlW1sf3cnORK4ATgviRLq2pTkqXA5l1suxZYC7BixYqanJwc6Zjnrrlml8tWL9/GhbfNNivOnnXMbR3rz5ocab2pqSlGfZ2Nk3VI0t5rjy8FJjkwyWO2TwPPAm4HrgbOaaudA1w12yIlSZIWg9l0I0wAVybZvp/3VNUHk3wKuDzJecDdwBmzL1OSJGnh2+NgVVVfAp68k/b7gZNnU5QkSdJi5DevS5IkdWKwkiRJ6mT+PzImzdKyaT4pOmz18m3Tfqp0R+svOG1PS5Ik7aPssZKkJsk7kmxOcvtQ22FJrkvyhfbz0NaeJG9Ksi7JrUmOn7/KJS0UBitJ+r53Aqfs0Lar8U+fAxzTHquAt81RjZIWMIOVJDVV9VHgazs072r805XAu2rgBuCQ9qXIkvZhBitJmt6uxj89ArhnaL0NrU3SPsyb1yVpRNONfzqdPR0bdVzjbC6EMSL35rEqPbfFqde5GawkaXq7Gv90I3DU0HpHtrYfsKdjo775kqvGMs7mqONrjtPePFal57Y49To3LwVK0vR2Nf7p1cDZ7dOBJwJbhi4ZStpH2WMlSU2SS4FJ4PAkG4BXAxew8/FPrwVOBdYBDwEvmvOCJS04BitJaqrqBbtY9APjn1ZVAeePtyJJi42XAiVJkjoxWEmSJHVisJIkSerEYCVJktSJwUqSJKkTg5UkSVInBitJkqRODFaSJEmdGKwkSZI6MVhJkiR1YrCSJEnqxGAlSZLUicFKkiSpE4OVJElSJwYrSZKkTpbMdwHSQrVszTVj2e87TzlwLPuVJM0/e6wkSZI6MVhJkiR14qVAaY7dtnEL547hMuP6C07rvk9J0u6xx0qSJKkTg5UkSVInBitJkqRODFaSJEmdGKwkSZI6MVhJkiR1YrCSJEnqxGAlSZLUydi+IDTJKcAbgf2At1fVBeM6liTNl8X4XjeucTD9klppTMEqyX7AW4BnAhuATyW5uqruHMfxJGk++F6nnXF0hX3buHqsTgDWVdWXAJJcBqwEfLORtDfxvW7I7vSErV6+beTwMa5AMa6eu9XLx7JbNeN63t55yoFd9jOuYHUEcM/Q/AbgqWM6liR2/81m1P/Y/Ct5Wr7XzYFx/UcqjUOqqv9Ok+cBp1TVS9r8C4GnVtVLh9ZZBaxqs08APt/h0IcDX+2wn9myjoezjofbm+s4uqoe23mfC9Yo73WtfU/f7xbKa2UcPLfFyXMb2OV73bh6rDYCRw3NH9navqeq1gJrex40yU1VtaLnPq3DOqxD05jxvQ72/P1ub36OPLfFyXOb2bi+buFTwDFJHp/kkcCZwNVjOpYkzRff6yQ9zFh6rKpqW5KXAn/D4CPI76iqO8ZxLEmaL77XSdrR2L7HqqquBa4d1/53oeulxVmwjoezjoezjr3ImN/r9ubnyHNbnDy3GYzl5nVJkqR9kUPaSJIkdbIog1WSo5J8JMmdSe5I8vLWfliS65J8of08dMx1PDrJJ5N8ttXxu6398UluTLIuyXvbTa1jl2S/JJ9J8oH5qiPJ+iS3JbklyU2tbU6fl3bMQ5JckeRzSe5K8rR5eH08of0etj++keQV8/T7+G/tNXp7kkvba3deXqd6uCSnJPl8ex7W7GT5o9rzs649X8vmvso9M8K5nZvkK0P/Rl4yH3XuiSTvSLI5ye27WJ4kb2rnfmuS4+e6xj01wrlNJtky9Lz9zlzXuKd2lR92WGd2z11VLboHsBQ4vk0/Bvh74FjgfwNrWvsa4A/GXEeAg9r0/sCNwInA5cCZrf1PgP86R7+X3wDeA3ygzc95HcB64PAd2ub0eWnHuRh4SZt+JHDIfNQxVM9+wL3A0fPwOj0C+AfggKHXxbnz9Tr18QOviy8CP95ep58Fjt1hnV8D/qRNnwm8d77r7nhu5wJ/PN+17uH5/QJwPHD7LpafCvx1+3/iRODG+a6547lNbv9/ZrE9dpUfej53i7LHqqo2VdWn2/SDwF0M/vNYyeA/VNrP08dcR1XV1ja7f3sU8AzgirmqAyDJkcBpwNvbfOajjl2Y0+clycEM3hguAqiqf6mqr891HTs4GfhiVd09T3UsAQ5IsgT4IWATC+f1sS/73pA4VfUvwPYhcYYNv16uAE5u/74XulHObdGqqo8CX5tmlZXAu9r/EzcAhyRZOjfVzc4I57ZoTZMfhs3quVuUwWpY6xZ/CoPeoomq2tQW3QtMzMHx90tyC7AZuI7BX2hfr6ptbZUN/OCTNg5vAH4L+G6b/5F5qqOADyW5OYNvm4a5f14eD3wF+LN2afTtSQ6chzqGnQlc2qbntI6q2gj8IfCPDALVFuBm5uf1oYfb2ZA4Oz4P31unPV9bGPz7XuhGOTeAf98ut1yR5KidLF+sRj3/xeppGdwG89dJfnq+i9kTO+SHYbN67hZ1sEpyEPA+4BVV9Y3hZTXozxv7Rx6r6jtVdRyDb1w+AXjiuI+5oyS/BNxpjvoAABl9SURBVGyuqpvn+tg78XNVdTzwHOD8JL8wvHCOnpclDLqx31ZVTwG+yeCS21zXAUC7d+m5wF/suGwu6mj3cK1kEDh/DDgQOGWcx5RG9FfAsqr6Nwz+ML14hvW1MHyawZAuTwbeDPzlPNez26bLD7O1aINVkv0Z/FIuqar3t+b7tnfXtZ+b56qedqnpI8DTGHQbbv+OsJ0OcdHZ04HnJlnPoLv9GcAb56GO7b0jVNVm4EoGYXOun5cNwIaq2v5XyBUMgtZ8vT6eA3y6qu5r83Ndx/8F/ENVfaWqvg28n8FrZs5fH/oBowyJ87112vN1MHD/nFQ3O6MMbXZ/VX2rzb4d+Jk5qm0ujDTc0WJUVd/YfhtMDb7Hbf8kh89zWSPbRX4YNqvnblEGq3Z/wUXAXVX1R0OLrgbOadPnAFeNuY7HJjmkTR8APJPB9dqPAM+bqzqq6pVVdWRVLWNwyenDVXXWXNeR5MAkj9k+DTwLuJ05fl6q6l7gniRPaE0nA3fOdR1DXsD3LwMyD3X8I3Bikh9q/3a2/z7m9PWhnRplSJzh18vzGPz7XgxfQDjjue1w38pzGbx/7i2uBs5unzA7EdgydAvAopbkR7ff55fkBAZZYjGE/enyw7DZPXfzdWf+bB7AzzG4fHIrcEt7nMrgvoPrgS8A/wc4bMx1/BvgM62O24Hfae0/DnwSWMfg8s+j5vB3M8n3PxU4p3W04322Pe4A/kdrn9PnpR3zOOCm9tz8JXDoPNVxIIM3nIOH2uajjt8FPtdep38OPGo+X6c+HvbcnMrgk0lfHPo383vAc9v0o9vzs649Xz8+3zV3PLf/u71XfJZB0H/ifNe8G+d2KYN7Fr/NoJf8POC/AP+lLQ/wlnbutwEr5rvmjuf20qHn7Qbg3853zbtxbrvKD92eO795XZIkqZNFeSlQkiRpITJYSZIkdWKwkiRJ6sRgJUmS1InBSpIk7RNmGmB6J+ufMTRg83tG2sZPBUqSpH1BGw1kK4OxAJ80w7rHMBis/hlV9UCSx9Xgy6+nZY+VJEnaJ9ROBphO8q+TfLCNcfuxJNuHpvvPwFuq6oG27UijZBisJEnSvmwt8LKq+hngN4G3tvafBH4yySeS3JBkpDFWl8y8iiRJ0t6nDcb8b4G/aKP0wGBkChhkpGMYjGhyJPDRJMtrMDbwLhmsJEnSvuoRwNer6ridLNsA3FiDwev/IcnfMwhan5pph5IkSfucqvoGg9D0fBgM0pzkyW3xXzLorSLJ4QwuDX5ppn0arCRJ0j4hyaXA3wFPSLIhyXnAWcB5ST7LYHDplW31vwHuT3Ing0HC/3tV3T/jMfy6BUmSpD7ssZIkSerEYCVJktSJwUqSJKkTg5UkSVInBitJkqRODFaSJEmdGKwkSZI6MVhJkiR1YrCSJEnqxGAlSZLUicFKkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdWKwkiRJ6sRgJUmS1InBSpIkqRODlSRJUicGK0mSpE4MVpIkSZ0YrCRJkjoxWEmSJHVisJIkSerEYKVFKclZST4033VIkjTMYLWAJXl3kj/boe0Xk9yfZOk81VRJvplk69Djt+a6jqq6pKqeNdfHlSRpOqmq+a5Bu5DkR4A7gBdW1XVJHg3cCvyvqnpnp2Msqaptu7F+AcdU1boex98Tu1uzJElzxR6rBayq7gdeBqxNciDwauCLVfXOJCcm+f+SfD3JZ5NMbt8uyYuS3JXkwSRfSvKrQ8smk2xI8ttJ7gX+LMnhST7Q9vW1JB9LstuvjSTXJrlwaP6yJO9o0+cm+USSP06yJcnnkpw8tO7BSS5KsinJxiSvTbLfDtu+Psn9wGta28eHtn9ikuta/Z9PcsbQsncmeUuSa9rv5MYk/3po+U8PbXtfkle19kckWZPki62X8PIkh+3u70WStO8wWC1wVfUXwKeBS4FVwKokRwDXAK8FDgN+E3hfkse2zTYDvwT8MPAi4PVJjh/a7Y+27Y5u+1wNbAAeC0wArwL2pCvzxcALkzwjyVnACcDLh5Y/FfgicDiDkPj+oaDyTmAb8BPAU4BnAS/ZYdsvtfpeN3zQFjqvA94DPA44E3hrkmOHVjsT+F3gUGDd9n0keQzwf4APAj/Wjn992+ZlwOnAL7ZlDwBv2b1fiSRpX2KwWhx+DXgG8HtVdQ/wn4Brq+raqvpuVV0H3AScClBV11TVF2vgb4EPAT8/tL/vAq+uqm9V1T8B3waWAkdX1ber6mM1/TXiT7fere2PZ7fj3gv8V+Bi4I3A2VX14NB2m4E3tGO8F/g8cFqSiVb7K6rqm1W1GXg9gzC03Zer6s1Vta3VPOyXgPVV9Wdt+WeA9wHPH1rnyqr6ZLuEeAlw3NC291bVhVX1z1X1YFXd2Jb9F+B/VNWGqvoW8BrgeUmWTPO7kSTtw/wPYhGoqvuSfJXB/VYw6Gl6fpJ/N7Ta/sBHAJI8h0GP0E8yCM8/BNw2tO5Xquqfh+b/Hwah4UNJANZW1QXTlHT8NPdY/RXwZuDzVfXxHZZt3CGw3c2gJ+joVv+mdnxa3fcMrTs8vaOjgacm+fpQ2xLgz4fm7x2afgg4qE0fxaAXbVf7vTLJd4favsOg12zjNPVIkvZRBqvF6R7gz6vqP++4IMmjGPTWnA1cVVXfTvKXQIZWe1hvVOtVWg2sTvIk4MNJPlVV17P7XgfcBTw+yQuq6tKhZUckyVC4+lfA1e18vgUcPs1N6dP1oN0D/G1VPXMP6r2Hh/eM7bjsxVX1iT3YryRpH+SlwMXp3cC/S/LsJPsleXS7Kf1I4JHAo4CvANta79W0X0uQ5JeS/EQG3UVbGPTKfHe6bXaxn19gcE/X2cA5wJvb/WDbPQ749ST7J3k+8FMMLmluYnC58sIkP9xuGv/XSX5xxEN/APjJJC9s+94/yc8m+akRt12a5BVJHpXkMUme2pb9CfC6JEe383tskpUj1iRJ2gcZrBahdp/VSgY3mX+FQc/Kfwce0Xqffh24nMHN1v+RQa/QdI5hcAP3VuDvgLdW1UemWf+zefj3WL0hyQ8D7wJeWlUbq+pjwEUMPnW4vbfsxnasrzLo2Xpe++QjDMLYI4E7W91XMLjva5Tfx4MMwuOZwJcZXPb7AwYBc5Rtnwn8u7bdF4CT2uI3MvjdfSjJg8ANDG6ilyRpp/weK82JJOcCL6mqn5vvWiRJGhd7rCRJkjoxWEmSJHXipUBJkqRO7LGSJEnqxGAlSZLUyYL4gtDDDz+8li1bNtK63/zmNznwwAPHW9AsWWMf1jh7c13fzTff/NWqeuzMa0rS3mlBBKtly5Zx0003jbTu1NQUk5OT4y1olqyxD2ucvbmuL8ndc3YwSVqAvBQoSZLUicFKkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdWKwkiRJ6sRgJUmS1InBSpIkqZMFMaSNFo5la64Z277XX3Da2PYtSdJCYI+VJElSJwYrSZKkTgxWkiRJnXiP1ZiN654l71eSJGnhscdKkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdWKwkiRJ6sRgJUmS1InBSpIkqRODlSRJUicGK0mSpE5mDFZJjkrykSR3Jrkjyctb+2FJrkvyhfbz0NaeJG9Ksi7JrUmOH/dJSJIkLQSj9FhtA1ZX1bHAicD5SY4F1gDXV9UxwPVtHuA5wDHtsQp4W/eqJUmSFqAZg1VVbaqqT7fpB4G7gCOAlcDFbbWLgdPb9ErgXTVwA3BIkqXdK5ckSVpgduseqyTLgKcANwITVbWpLboXmGjTRwD3DG22obVJkiTt1VJVo62YHAT8LfC6qnp/kq9X1SFDyx+oqkOTfAC4oKo+3tqvB367qm7aYX+rGFwqZGJi4mcuu+yykerYunUrBx100EjrzpfhGm/buGUsx1h+xMGz2n5Xv8dx1Qu7X/Nie64Xormu76STTrq5qlbM2QElaYFZMspKSfYH3gdcUlXvb833JVlaVZvapb7NrX0jcNTQ5ke2toepqrXAWoAVK1bU5OTkSAVPTU0x6rrzZbjGc9dcM5ZjrD9rclbb7+r3OK56YfdrXmzP9UK00OuTpL3NKJ8KDHARcFdV/dHQoquBc9r0OcBVQ+1nt08HnghsGbpkKEmStNcapcfq6cALgduS3NLaXgVcAFye5DzgbuCMtuxa4FRgHfAQ8KKuFUuSJC1QMwardq9UdrH45J2sX8D5s6xLkiRp0fGb1yVJkjoxWEmSJHVisJIkSerEYCVJktSJwUqSJKkTg5UkSVInBitJkqRODFaSJEmdGKwkSZI6MVhJkiR1YrCSJEnqxGAlSZLUicFKkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdWKwkiRJ6sRgJUmS1InBSpIkqRODlSRJUicGK0mSpE4MVpIkSZ0YrCRJkjoxWEmSJHVisJIkSerEYCVJktSJwUqSJKkTg5UkSVInBitJkqRODFaSJEmdGKwkSZI6MVhJkiR1YrCSJEnqxGAlSZLUicFKkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdTJjsEryjiSbk9w+1PaaJBuT3NIepw4te2WSdUk+n+TZ4ypckiRpoRmlx+qdwCk7aX99VR3XHtcCJDkWOBP46bbNW5Ps16tYSZKkhWzGYFVVHwW+NuL+VgKXVdW3quofgHXACbOoT5IkadGYzT1WL01ya7tUeGhrOwK4Z2idDa1NkiRpr5eqmnmlZBnwgap6UpufAL4KFPD7wNKqenGSPwZuqKp3t/UuAv66qq7YyT5XAasAJiYmfuayyy4bqeCtW7dy0EEHjbTufBmu8baNW8ZyjOVHHDyr7Xf1exxXvbD7NS+253ohmuv6TjrppJurasWcHVCSFpgle7JRVd23fTrJnwIfaLMbgaOGVj2yte1sH2uBtQArVqyoycnJkY49NTXFqOvOl+Eaz11zzViOsf6syVltv6vf47jqhd2vebE91wvRQq9PkvY2e3QpMMnSodlfBrZ/YvBq4Mwkj0ryeOAY4JOzK1GSJGlxmLHHKsmlwCRweJINwKuBySTHMbgUuB74VYCquiPJ5cCdwDbg/Kr6znhK12KzbDd7w1Yv3zZSD9r6C07b05IkSepqxmBVVS/YSfNF06z/OuB1sylKkiRpMfKb1yVJkjoxWEmSJHVisJIkSerEYCVJktSJwUqSJKkTg5UkSVInBitJkqRODFaSJEmdGKwkSZI6MVhJkiR1YrCSJEnqxGAlSZLUyYyDMGthWrbmmlltv3r5Ns6d5T4kSdLD2WMlSZLUicFKkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdWKwkiRJ6sRgJUmS1InBSpIkqRODlSRJUicGK0mSpE4MVpIkSZ0YrCRJkjoxWEmSJHWyZL4LWCiWrbmm275WL9/GuR33J0mSFgd7rCRJkjoxWEmSJHVisJIkSerEYCVJktSJwUqSJKkTg5UkSVInBitJkqRODFaSJEmdGKwkSZI6MVhJkiR1YrCSJEnqxGAlSZLUicFKkiSpkxmDVZJ3JNmc5PahtsOSXJfkC+3noa09Sd6UZF2SW5McP87iJUmSFpJReqzeCZyyQ9sa4PqqOga4vs0DPAc4pj1WAW/rU6YkSdLCN2OwqqqPAl/boXklcHGbvhg4faj9XTVwA3BIkqW9ipUkSVrI9vQeq4mq2tSm7wUm2vQRwD1D621obZIkSXu9JbPdQVVVktrd7ZKsYnC5kImJCaampkbabuvWrSOvuztWL9/WbV8TB/Td3zjsTTWO4/UwqnG9HntZ6PVJ0t5mT4PVfUmWVtWmdqlvc2vfCBw1tN6Rre0HVNVaYC3AihUranJycqQDT01NMeq6u+PcNdd029fq5du48LZZZ9ax2ptqXH/W5PiL2YVxvR57Wej1SdLeZk8vBV4NnNOmzwGuGmo/u3068ERgy9AlQ0mSpL3ajN0BSS4FJoHDk2wAXg1cAFye5DzgbuCMtvq1wKnAOuAh4EVjqFmSJGlBmjFYVdULdrHo5J2sW8D5sy1KkiRpMfKb1yVJkjoxWEmSJHVisJIkSerEYCVJktSJwUqSJKkTg5UkSVInBitJkqRODFaSJEmdGKwkSZI6MVhJkiR1YrCSJEnqxGAlSZLUicFKkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdWKwkiRJ6sRgJUmS1InBSpIkqRODlSRJUicGK0mSpE4MVpIkSZ0YrCRJkjoxWEmSJHVisJIkSepkyXwXsLtu27iFc9dcM99lSJIk/QB7rCRJkjoxWEmSJHVisJIkSerEYCVJktSJwUqSJKkTg5UkSVInBitJkqRODFaSJEmdGKwkSZI6MVhJkiR1YrCSJEnqxGAlSZLUicFKkiSpE4OVJElSJ0vmuwBptpatuWYs+11/wWlj2a8kae81q2CVZD3wIPAdYFtVrUhyGPBeYBmwHjijqh6YXZmSJEkLX49LgSdV1XFVtaLNrwGur6pjgOvbvCRJ0l5vHPdYrQQubtMXA6eP4RiSJEkLzmyDVQEfSnJzklWtbaKqNrXpe4GJWR5DkiRpUUhV7fnGyRFVtTHJ44DrgJcBV1fVIUPrPFBVh+5k21XAKoCJiYmfueyyy0Y65uavbeG+f9rjkufExAFYYwfzXePyIw6ecZ2tW7dy0EEHzUE1e2au6zvppJNuHrotQJL2ObO6eb2qNrafm5NcCZwA3JdkaVVtSrIU2LyLbdcCawFWrFhRk5OTIx3zzZdcxYW3LewPM65evs0aO5jvGtefNTnjOlNTU4z62p0PC70+Sdrb7PGlwCQHJnnM9mngWcDtwNXAOW21c4CrZlukJEnSYjCb7oAJ4Mok2/fznqr6YJJPAZcnOQ+4Gzhj9mVKkiQtfHscrKrqS8CTd9J+P3DybIqSJElajBzSRpIkqRODlSRJUicGK0mSpE4MVpIkSZ0YrCRJkjoxWEmSJHVisJIkSerEYCVJktSJwUqSJKkTg5UkSVInBitJkqRODFaSJEmdGKwkSZI6MVhJkiR1smS+C5AWqmVrrplxndXLt3HuCOsNW3/BaXtakiRpgbPHSpIkqRODlSRJUicGK0mSpE4MVpIkSZ0YrCRJkjoxWEmSJHVisJIkSerEYCVJktSJwUqSJKkTg5UkSVInBitJkqRODFaSJEmdGKwkSZI6MVhJkiR1YrCSJEnqxGAlSZLUicFKkiSpE4OVJElSJwYrSZKkTgxWkiRJnRisJEmSOjFYSZIkdWKwkiRJ6mTJfBcg7WuWrblmLPtdf8FpY9mvJGl09lhJkiR1YrCSJEnqxGAlSZLUydiCVZJTknw+yboka8Z1HEmSpIViLMEqyX7AW4DnAMcCL0hy7DiOJUmStFCMq8fqBGBdVX2pqv4FuAxYOaZjSZIkLQjj+rqFI4B7huY3AE8d07EksfOvcVi9fBvnzvLrHfwaB0kaXaqq/06T5wGnVNVL2vwLgadW1UuH1lkFrGqzTwA+P+LuDwe+2rHccbDGPqxx9ua6vqOr6rFzeDxJWlDG1WO1EThqaP7I1vY9VbUWWLu7O05yU1WtmF1542WNfVjj7C30+iRpbzOue6w+BRyT5PFJHgmcCVw9pmNJkiQtCGPpsaqqbUleCvwNsB/wjqq6YxzHkiRJWijGNlZgVV0LXDuGXe/25cN5YI19WOPsLfT6JGmvMpab1yVJkvZFDmkjSZLUyaIKVothmJwk65PcluSWJDfNdz0ASd6RZHOS24faDktyXZIvtJ+HLsAaX5NkY/td3pLk1Hms76gkH0lyZ5I7kry8tS+Y3+M0NS6Y36Mk7e0WzaXANkzO3wPPZPCFo58CXlBVd85rYTtIsh5YUVUL5ruNkvwCsBV4V1U9qbX9b+BrVXVBC6mHVtVvL7AaXwNsrao/nK+6tkuyFFhaVZ9O8hjgZuB0+P/bu18XK8IoDuPPCaY1CttE9B9Q84ZNgkksC6a1GTSYLSajYjOIYYMKgj+jUZNBgwpbNch1Nxg0u8cw7+J1vXsFGXbOXJ5PucMMF76cMoc578zLRYrUcU7GNYrUUZIW3ZieWLlNzn/KzFfAtz2nzwEb7XiD7gY8mH0ylpGZk8x8145/AJt0OwyUqeOcjJKkAzKmxmrWNjkVbxoJvIyIt+3r8lUtZ+akHX8FlocMM8eViHjfRoWDjit3RcQx4BTwhqJ13JMRCtZRkhbRmBqrsVjJzNPAWeByG3GVlt08uOJM+A5wAjgJTICbw8aBiDgMPAauZub36WtV6jgjY7k6StKiGlNj9c9tcirIzC/tdxt4SjfCrGirrcnZXZuzPXCev2TmVmb+zMwd4C4D1zIiDtE1LPcz80k7XaqOszJWq6MkLbIxNVblt8mJiKW2aJiIWALOAB/n/2swL4D1drwOPB8wy0y7DUtzngFrGREB3AM2M/PW1KUyddwvY6U6StKiG81bgQDtNfHb/N4m58bAkf4QEcfpnlJB91X7BxUyRsRDYBU4AmwB14FnwCPgKPAZWMvMwRaP75NxlW58lcAn4NLUeqaDzrcCvAY+ADvt9DW6NUwl6jgn4wWK1FGSFt2oGitJkqTKxjQKlCRJKs3GSpIkqSc2VpIkST2xsZIkSeqJjZUkSVJPbKwkSZJ6YmMlSZLUExsrSZKknvwCoHo2RH4TE2YAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 720x720 with 4 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JjhHcqbmUBWI",
        "colab_type": "code",
        "outputId": "c05b582a-0511-41dd-901e-356671c8e6dd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        }
      },
      "source": [
        "so_survey_df[['Age', 'Years Experience']].boxplot()\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TmM2zW8QUYTf",
        "colab_type": "code",
        "outputId": "34f1a89a-f0ab-4ac8-b5fc-9504dd43faa6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 276
        }
      },
      "source": [
        "so_survey_df[['ConvertedSalary']].boxplot()\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YdADha5dUocv",
        "colab_type": "code",
        "outputId": "8ed23c39-0d23-44b3-d0d4-9e8bb6b81978",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 578
        }
      },
      "source": [
        "sns.pairplot(so_numeric_df)\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 540x540 with 12 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "op7uFM-70CW5",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Normalization\n",
        "---\n",
        "In normalization we linearly scale the entire column between 0 and 1, with 0 corresponding with the lowest value in the column, and 1 with the largest. When using scikit-learn (the most commonly used machine learning library in Python) we can use a `MinMaxScaler` to apply normalization. (It is called this as it scales our values between a minimum and maximum value.)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ME9djjbu0K83",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Fit the `MinMaxScaler` on the `Age` column of `so_numeric_df`.\n",
        "* Transform the same column with the scaler you just fit."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jNgwBJRp0Zgl",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Import MinMaxScaler\n",
        "from sklearn.preprocessing import MinMaxScaler"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iYeEJLVQ0HCU",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Instantiate MinMaxScaler\n",
        "MM_scaler = MinMaxScaler()\n",
        "MM_scaler.fit(so_numeric_df[['Age']])\n",
        "so_numeric_df['Age_MM'] = MM_scaler.transform(so_numeric_df[['Age']])\n",
        "print(so_survey_df[['Age_MM', 'Age']].head())"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lBKJrNv60m2p",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Standardization\n",
        "---\n",
        "While normalization can be useful for scaling a column between two data points, it is hard to compare two scaled columns if even one of them is overly affected by outliers. One commonly used solution to this is called standardization, where instead of having a strict upper and lower bound, you center the data around its mean, and calculate the number of standard deviations away from mean each data point is."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gFyXLW3v0rU9",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Fit the `StandardScaler` on the `Age` column of `so_numeric_df`.\n",
        "* Transform the same column with the scaler we just fit."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TL8N_V9n03jy",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Import StandardScaler\n",
        "from sklearn.preprocessing import StandardScaler"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lOoRs_vR0qmJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Instantiate StandardScaler\n",
        "SS_scaler = StandardScaler()\n",
        "SS_scaler.fit(so_numeric_df[['Age']])\n",
        "so_numeric_df['Age_SS'] = SS_scaler.transform(so_numeric_df[['Age']])\n",
        "print(so_survey_df[['Age_SS', 'Age']].head())"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QLntVWcp1aZ1",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Log transformation\n",
        "---\n",
        "In the previous exercises we scaled the data linearly, which will not affect the data's shape. This works great if our data is normally distributed (or closely normally distributed), an assumption that a lot of machine learning models make. Sometimes we will work with data that closely conforms to normality, e.g the height or weight of a population. On the other hand, many variables in the real world do not follow this pattern e.g, wages or age of a population. \n",
        "\n",
        "Now, we will use a log transform on the `ConvertedSalary` column in the `so_numeric_df` DataFrame as it has a large amount of its data centered around the lower values, but contains very high values also. These distributions are said to have a long right tail."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NMQB-F4e1lMK",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Fit the `PowerTransformer` on the `ConvertedSalary` column of `so_numeric_df`.\n",
        "* Transform the same column with the scaler we just fit."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "N95T8tjM1h5A",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Import PowerTransformer\n",
        "from sklearn.preprocessing import PowerTransformer"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9CNzVVOg1ueY",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Instantiate PowerTransformer\n",
        "pow_trans = PowerTransformer()\n",
        "pow_trans.fit(so_numeric_df[['ConvertedSalary']])\n",
        "so_numeric_df['ConvertedSalary_1'] = pow_trans.transform(so_numeric_df[['ConvertedSalary']])\n",
        "so_survey_df[['ConvertedSalary_1', 'ConvertedSalary']].hist()\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VhlDOrtb18uG",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Percentage based outlier removal\n",
        "---\n",
        "One way to ensure a small portion of data is not having an overly adverse effect is by removing a certain percentage of the largest and/or smallest values in the column. This can be achieved by finding the relevant quantile and trimming the data using it with a mask. This approach is particularly useful if we are concerned that the highest values in our dataset should be avoided. When using this approach, we must remember that even if there are no outliers, this will still remove the same top N percentage from the dataset."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AhiOqx-v2CFB",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Find the 95th quantile of the `ConvertedSalary` column in `so_numeric_df` DataFrame.\n",
        "* Trim the `so_numeric_df` DataFrame to retain all rows where `ConvertedSalary` is less than it's 95th quantile and store this as `trimmed_df`.\n",
        "* Plot the histogram of `so_numeric_df[['ConvertedSalary']]`.\n",
        "* Plot the histogram of `trimmed_df[['ConvertedSalary']]`"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "G1_-eGVu1_JA",
        "colab_type": "code",
        "outputId": "61ab2086-e181-44e2-cb72-3168e7c5ea79",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 575
        }
      },
      "source": [
        "# Find the 95th quantile\n",
        "quantile = so_numeric_df['ConvertedSalary'].quantile(0.95)\n",
        "trimmed_df= so_numeric_df[so_numeric_df['ConvertedSalary']<quantile]\n",
        "so_numeric_df[['ConvertedSalary']].hist()\n",
        "plt.show()\n",
        "plt.clf\n",
        "\n",
        "trimmed_df[['ConvertedSalary']].hist()\n",
        "plt.show()\n",
        "plt.clf"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<function matplotlib.pyplot.clf>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 38
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Lgva617FNqZy",
        "colab_type": "text"
      },
      "source": [
        "## Dealing with Text Data\n",
        "---"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IuHUqVR03PJV",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Cleaning up your text\n",
        "---\n",
        "Unstructured text data cannot be directly used in most analyses. Multiple steps need to be taken to go from a long free form string to a set of numeric columns in the right format that can be ingested by a machine learning model. The first step of this process is to standardize the data and eliminate any characters that could cause problems later on in your analytic pipeline.\n",
        "\n",
        "Here, we will be working with a new dataset containing the inaugural speeches of the presidents of the United States loaded as `speech_df`, with the speeches stored in the `text` column."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dFmYNs7W3hYj",
        "colab_type": "code",
        "outputId": "f5d09df6-a509-40da-ea77-78ba33fcd525",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        }
      },
      "source": [
        "speech_df = pd.read_csv('https://raw.githubusercontent.com/shala2020/shala2020.github.io/master/Lecture_Materials/Assignments/MachineLearning/L6/inaugural_speeches.csv')\n",
        "speech_df.head()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "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>Inaugural Address</th>\n",
              "      <th>Date</th>\n",
              "      <th>text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>George Washington</td>\n",
              "      <td>First Inaugural Address</td>\n",
              "      <td>Thursday, April 30, 1789</td>\n",
              "      <td>Fellow-Citizens of the Senate and of the House...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>George Washington</td>\n",
              "      <td>Second Inaugural Address</td>\n",
              "      <td>Monday, March 4, 1793</td>\n",
              "      <td>Fellow Citizens:  I AM again called upon by th...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>John Adams</td>\n",
              "      <td>Inaugural Address</td>\n",
              "      <td>Saturday, March 4, 1797</td>\n",
              "      <td>WHEN it was first perceived, in early times, t...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Thomas Jefferson</td>\n",
              "      <td>First Inaugural Address</td>\n",
              "      <td>Wednesday, March 4, 1801</td>\n",
              "      <td>Friends and Fellow-Citizens:  CALLED upon to u...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Thomas Jefferson</td>\n",
              "      <td>Second Inaugural Address</td>\n",
              "      <td>Monday, March 4, 1805</td>\n",
              "      <td>PROCEEDING, fellow-citizens, to that qualifica...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                Name  ...                                               text\n",
              "0  George Washington  ...  Fellow-Citizens of the Senate and of the House...\n",
              "1  George Washington  ...  Fellow Citizens:  I AM again called upon by th...\n",
              "2         John Adams  ...  WHEN it was first perceived, in early times, t...\n",
              "3   Thomas Jefferson  ...  Friends and Fellow-Citizens:  CALLED upon to u...\n",
              "4   Thomas Jefferson  ...  PROCEEDING, fellow-citizens, to that qualifica...\n",
              "\n",
              "[5 rows x 4 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 39
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qvbCog623SDx",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Print the first 5 rows of the `text` column in `speech_df` DataFrame to see the free text fields.\n",
        "* Replace all non letter characters in the `text` column with a whitespace and add it as a new column `text_clean` in the `speech_df` DataFrame. \n",
        "* Make all characters in the newly created `text_clean` column lower case.\n",
        "* Print the first 5 rows of the `text_clean` column. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ug22N0u2edBr",
        "colab_type": "code",
        "outputId": "d62a0e68-5e1a-4fa0-bbbb-65771faf6cca",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "print(speech_df.text.head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0    Fellow-Citizens of the Senate and of the House...\n",
            "1    Fellow Citizens:  I AM again called upon by th...\n",
            "2    WHEN it was first perceived, in early times, t...\n",
            "3    Friends and Fellow-Citizens:  CALLED upon to u...\n",
            "4    PROCEEDING, fellow-citizens, to that qualifica...\n",
            "Name: text, dtype: object\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7qIvRoX9fJqT",
        "colab_type": "code",
        "outputId": "be6d7e2d-b223-47be-c20a-2759e2e3c747",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "speech_df['text_clean']=speech_df['text'].str.replace('[^a-zA-Z]',' ')\n",
        "speech_df['text_clean']=speech_df['text_clean'].str.lower()\n",
        "print(speech_df['text_clean'].head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0    fellow citizens of the senate and of the house...\n",
            "1    fellow citizens   i am again called upon by th...\n",
            "2    when it was first perceived  in early times  t...\n",
            "3    friends and fellow citizens   called upon to u...\n",
            "4    proceeding  fellow citizens  to that qualifica...\n",
            "Name: text_clean, dtype: object\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ag6i4ijT4tcM",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### High level text features\n",
        "---\n",
        "Once the text has been cleaned and standardized we can begin creating features from the data. The most fundamental information we can calculate about free form text is its size, such as its length and number of words. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "q4jPqRmA4zRA",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Record the character length of each speech (`speech_df['text_clean']`) and store it in a new `char_count` column.\n",
        "* Record the word count of each speech in the `word_count` column.\n",
        "* Record the average word length of each speech in the `avg_word_length` column.\n",
        "* Print the first 5 rows of the columns: `text_clean`, `char_cnt`, `word_cnt`, `avg_word_length`\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HgMZD5eHeeQB",
        "colab_type": "code",
        "outputId": "3116109a-3585-40d9-c2c4-5e35fcb1abb8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 153
        }
      },
      "source": [
        "speech_df['char_count'] = speech_df['text_clean'].str.len()\n",
        "speech_df['word_count'] = speech_df['text_clean'].str.split().str.len()\n",
        "speech_df['avg_word_length'] = speech_df['char_count'] / speech_df['word_count']\n",
        "print(speech_df[['text_clean', 'char_count', 'word_count', 'avg_word_length']].head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "                                          text_clean  ...  avg_word_length\n",
            "0  fellow citizens of the senate and of the house...  ...         6.016760\n",
            "1  fellow citizens   i am again called upon by th...  ...         5.829630\n",
            "2  when it was first perceived  in early times  t...  ...         5.971158\n",
            "3  friends and fellow citizens   called upon to u...  ...         5.843318\n",
            "4  proceeding  fellow citizens  to that qualifica...  ...         5.948363\n",
            "\n",
            "[5 rows x 4 columns]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Wi2hOh045bsm",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Counting words (I)\n",
        "---\n",
        "Once high level information has been recorded we can begin creating features based on the actual content of each text, as given below: \n",
        "\n",
        "* For each unique word in the dataset a column is created.\n",
        "* For each entry, the number of times this word occurs is counted and the count value is entered into the respective column.\n",
        "\n",
        "These \"count\" columns can then be used to train machine learning models."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "b6sMPuHi5nB7",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Import `CountVectorizer` from `sklearn.feature_extraction.text`.\n",
        "* Instantiate `CountVectorizer` and assign it to 'cv'.\n",
        "* Fit the vectorizer to the `text_clean` column.\n",
        "* Print the feature names generated by the vectorizer and find the number of features.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nomeBUIc5lA2",
        "colab_type": "code",
        "outputId": "0b687dc7-6b57-4dbc-8673-7c5d983299bc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# Import CountVectorizer\n",
        "from sklearn.feature_extraction.text import CountVectorizer\n",
        "\n",
        "# Instantiate CountVectorizer\n",
        "cv = CountVectorizer()\n",
        "cv.fit(speech_df['text_clean'])\n",
        "print(len(cv.get_feature_names()))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "9043\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bAm6_JHr6Cto",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Counting words (II)\n",
        "---\n",
        "Once the vectorizer has been fit to the data, it can be used to transform the text to an array representing the word counts. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JcFMh9k_6Hu4",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Apply the vectorizer ('cv' in the previous exercise) to the `text_clean` column.\n",
        "* Convert this transformed (sparse) array into a `numpy` array with counts and print it. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zu-Uvsjki-2i",
        "colab_type": "code",
        "outputId": "f181b931-831e-49ec-c9d3-c7e4015d7fea",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 136
        }
      },
      "source": [
        "cv_transformed = cv.transform(speech_df['text_clean'])\n",
        "cv_array = cv_transformed.toarray()\n",
        "print(cv_array)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[0 0 0 ... 0 0 0]\n",
            " [0 0 0 ... 0 0 0]\n",
            " [0 1 0 ... 0 0 0]\n",
            " ...\n",
            " [0 1 0 ... 0 0 0]\n",
            " [0 0 0 ... 0 0 0]\n",
            " [0 0 0 ... 0 0 0]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vUQAwgOV73BB",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Limiting your features\n",
        "---\n",
        "As we have seen, using the `CountVectorizer` with its default settings creates a feature for every single word in our corpus. This can create far too many features, often including ones that will provide very little analytical value.\n",
        "\n",
        "For this purpose `CountVectorizer` has parameters that you can set to reduce the number of features:\n",
        "\n",
        "* `min_df` : Use only words that occur in more than this percentage of documents. This can be used to remove outlier words that will not generalize across texts.\n",
        "* `max_df` : Use only words that occur in less than this percentage of documents. This is useful to eliminate very common words that occur in every corpus without adding value such as \"and\" or \"the\"."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "I9X2x50g8Dfd",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Limit the number of features in the `CountVectorizer` by setting the minimum number of documents a word can appear to 20% and the maximum to 80%.\n",
        "* Fit and apply the vectorizer on `text_clean` column in one step.\n",
        "* Convert this transformed (sparse) array into a `numpy` array with counts and print the dimensions of the new reduced array.\n",
        "* Did you notice that the number of features (unique words) greatly reduced from 9043 to 818?\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RqNCtZu-8NbI",
        "colab_type": "code",
        "outputId": "d6471d95-b8ee-4e9f-c71c-cf7479b629b1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# Import CountVectorizer\n",
        "from sklearn.feature_extraction.text import CountVectorizer\n",
        "\n",
        "# Specify arguements to limit the number of features generated\n",
        "cv = CountVectorizer(min_df=0.2, max_df=0.8)\n",
        "\n",
        "cv_transformed= cv.fit_transform(speech_df['text_clean'])\n",
        "cv_array = cv_transformed.toarray()\n",
        "print(cv_array.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(58, 818)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KKZ4xX3L9IxL",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Text to DataFrame\n",
        "---\n",
        "Now that we have generated these count based features in an array we will need to reformat them so that they can be combined with the rest of the dataset. This can be achieved by converting the array into a pandas DataFrame, with the feature names you found earlier as the column names, and then concatenate it with the original DataFrame.\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rvioGe8x9Olj",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Create a DataFrame `cv_df` containing the `cv_array` as the values and the feature names as the column names.\n",
        "* Add the prefix `Counts_` to the column names for ease of identification.\n",
        "* Concatenate this DataFrame (`cv_df`) to the original DataFrame (`speech_df`) column wise.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2lqa9WR_jsVA",
        "colab_type": "code",
        "outputId": "698ef966-a513-4652-c9e2-bde166d70472",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 153
        }
      },
      "source": [
        "cv_df =pd.DataFrame(cv_array, columns=cv.get_feature_names()).add_prefix('Counts_')\n",
        "speech_df_new=pd.concat([speech_df, cv_df], axis=1, sort=False)\n",
        "print(speech_df_new.head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "                Name         Inaugural Address  ... Counts_young Counts_your\n",
            "0  George Washington   First Inaugural Address  ...            0           9\n",
            "1  George Washington  Second Inaugural Address  ...            0           1\n",
            "2         John Adams         Inaugural Address  ...            0           1\n",
            "3   Thomas Jefferson   First Inaugural Address  ...            0           7\n",
            "4   Thomas Jefferson  Second Inaugural Address  ...            0           4\n",
            "\n",
            "[5 rows x 826 columns]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Rpo0ek00-dsr",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Tf-idf\n",
        "---\n",
        "While counts of occurrences of words can be useful to build models, words that occur many times may skew the results undesirably. To limit these common words from overpowering your model a form of normalization can be used. In this lesson we will be using **Term frequency-inverse document frequency** (**Tf-idf**). Tf-idf has the effect of reducing the value of common words, while increasing the weight of words that do not occur in many documents."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "c29xSq9Q-mcH",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Import `TfidfVectorizer` from `sklearn.feature_extraction.text`.\n",
        "* Instantiate `TfidfVectorizer` while limiting the number of features to 100 and removing English stop words.\n",
        "* Fit and apply the vectorizer on `text_clean` column in one step.\n",
        "* Create a DataFrame `tv_df` containing the weights of the words and the feature names as the column names.\n",
        "* Add the prefix `TFIDF_` to the column names for ease of identification."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-F0vdfx9-yIq",
        "colab_type": "code",
        "outputId": "a6c8c750-3c88-4186-cdde-05986917059b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 153
        }
      },
      "source": [
        "# Import TfidfVectorizer\n",
        "from sklearn.feature_extraction.text import TfidfVectorizer\n",
        "\n",
        "# Instantiate TfidfVectorizer\n",
        "tv = TfidfVectorizer(max_features=100, stop_words='english')\n",
        "tv_transformed = tv.fit_transform(speech_df['text_clean'])\n",
        "tv_df = pd.DataFrame(tv_transformed.toarray(), columns=tv.get_feature_names()).add_prefix('TFIDF_')\n",
        "print(tv_df.head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "   TFIDF_action  TFIDF_administration  ...  TFIDF_world  TFIDF_years\n",
            "0      0.000000              0.133415  ...     0.045929     0.052694\n",
            "1      0.000000              0.261016  ...     0.000000     0.000000\n",
            "2      0.000000              0.092436  ...     0.063643     0.073018\n",
            "3      0.000000              0.092693  ...     0.095729     0.000000\n",
            "4      0.041334              0.039761  ...     0.054752     0.062817\n",
            "\n",
            "[5 rows x 100 columns]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qzcGgNmS_Yc5",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Inspecting Tf-idf values\n",
        "---\n",
        "After creating Tf-idf features we will often want to understand what are the most highest scored words for each corpus. This can be achieved by isolating the row we want to examine and then sorting the the scores from high to low."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MP0QjLZJ_cvX",
        "colab_type": "text"
      },
      "source": [
        "* Assign the first row of `tv_df` to `sample_row`.\n",
        "* `sample_row` is now a series of weights assigned to words. Sort these values to print the top 5 highest-rated words."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g800sJg2_j2x",
        "colab_type": "code",
        "outputId": "b3972a50-1ee3-4918-996a-41e18253234e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "# Isolate the row to be examined\n",
        "sample_row = tv_df.iloc[0]\n",
        "print(sample_row.sort_values(ascending=False).head())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "TFIDF_government    0.367430\n",
            "TFIDF_public        0.333237\n",
            "TFIDF_present       0.315182\n",
            "TFIDF_duty          0.238637\n",
            "TFIDF_citizens      0.229644\n",
            "Name: 0, dtype: float64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "A7WK0CHgH_3J",
        "colab_type": "text"
      },
      "source": [
        "---\n",
        "### Sentiment analysis\n",
        "---\n",
        "You have been given the tweets about US airlines. Making use of this data, your task is to predict whether a tweet contains \n",
        "\n",
        "* positive, \n",
        "* negative, or \n",
        "* neutral sentiment \n",
        "\n",
        "about the airline."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mupsP8kXIFld",
        "colab_type": "code",
        "outputId": "6104534d-5ab3-4fe5-d55c-1294b6dbf4ac",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 445
        }
      },
      "source": [
        "data_source_url = \"https://raw.githubusercontent.com/shala2020/shala2020.github.io/master/Lecture_Materials/Assignments/MachineLearning/L6/Tweets.csv\"\n",
        "airline_tweets = pd.read_csv(data_source_url)\n",
        "airline_tweets.head()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "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>tweet_id</th>\n",
              "      <th>airline_sentiment</th>\n",
              "      <th>airline_sentiment_confidence</th>\n",
              "      <th>negativereason</th>\n",
              "      <th>negativereason_confidence</th>\n",
              "      <th>airline</th>\n",
              "      <th>airline_sentiment_gold</th>\n",
              "      <th>name</th>\n",
              "      <th>negativereason_gold</th>\n",
              "      <th>retweet_count</th>\n",
              "      <th>text</th>\n",
              "      <th>tweet_coord</th>\n",
              "      <th>tweet_created</th>\n",
              "      <th>tweet_location</th>\n",
              "      <th>user_timezone</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>570306133677760513</td>\n",
              "      <td>neutral</td>\n",
              "      <td>1.0000</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Virgin America</td>\n",
              "      <td>NaN</td>\n",
              "      <td>cairdin</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0</td>\n",
              "      <td>@VirginAmerica What @dhepburn said.</td>\n",
              "      <td>NaN</td>\n",
              "      <td>2015-02-24 11:35:52 -0800</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Eastern Time (US &amp; Canada)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>570301130888122368</td>\n",
              "      <td>positive</td>\n",
              "      <td>0.3486</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0.0000</td>\n",
              "      <td>Virgin America</td>\n",
              "      <td>NaN</td>\n",
              "      <td>jnardino</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0</td>\n",
              "      <td>@VirginAmerica plus you've added commercials t...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>2015-02-24 11:15:59 -0800</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Pacific Time (US &amp; Canada)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>570301083672813571</td>\n",
              "      <td>neutral</td>\n",
              "      <td>0.6837</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Virgin America</td>\n",
              "      <td>NaN</td>\n",
              "      <td>yvonnalynn</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0</td>\n",
              "      <td>@VirginAmerica I didn't today... Must mean I n...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>2015-02-24 11:15:48 -0800</td>\n",
              "      <td>Lets Play</td>\n",
              "      <td>Central Time (US &amp; Canada)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>570301031407624196</td>\n",
              "      <td>negative</td>\n",
              "      <td>1.0000</td>\n",
              "      <td>Bad Flight</td>\n",
              "      <td>0.7033</td>\n",
              "      <td>Virgin America</td>\n",
              "      <td>NaN</td>\n",
              "      <td>jnardino</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0</td>\n",
              "      <td>@VirginAmerica it's really aggressive to blast...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>2015-02-24 11:15:36 -0800</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Pacific Time (US &amp; Canada)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>570300817074462722</td>\n",
              "      <td>negative</td>\n",
              "      <td>1.0000</td>\n",
              "      <td>Can't Tell</td>\n",
              "      <td>1.0000</td>\n",
              "      <td>Virgin America</td>\n",
              "      <td>NaN</td>\n",
              "      <td>jnardino</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0</td>\n",
              "      <td>@VirginAmerica and it's a really big bad thing...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>2015-02-24 11:14:45 -0800</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Pacific Time (US &amp; Canada)</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "             tweet_id  ...               user_timezone\n",
              "0  570306133677760513  ...  Eastern Time (US & Canada)\n",
              "1  570301130888122368  ...  Pacific Time (US & Canada)\n",
              "2  570301083672813571  ...  Central Time (US & Canada)\n",
              "3  570301031407624196  ...  Pacific Time (US & Canada)\n",
              "4  570300817074462722  ...  Pacific Time (US & Canada)\n",
              "\n",
              "[5 rows x 15 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 51
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AYlKJmRzIHlu",
        "colab_type": "text"
      },
      "source": [
        "Instructions:\n",
        "\n",
        "* Draw a pie chart to see the number of tweets for each airline. \n",
        "* Draw a pie chart to see the distribution of sentiments across all the tweets. \n",
        "* Apply suitable data pre-processing steps to get rid of undesired symbols. This part has been done upto some extent.  \n",
        "* Using `TfidfVectorizer` class, convert text features into TF-IDF feature vectors.\n",
        "* `airline_sentiment` is the label and `text` is the feature. Apply suitable `train_test_split`, implement suitable machine learning classifier, and show the accuracy. \n",
        "\n",
        "Note: \n",
        "\n",
        "* For your convenience, the pre-processing of features have been done. \n",
        "* You are supposed to apply the vectorization on `processed_features` and implement ML algorithm. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SmFNdV_QnZHn",
        "colab_type": "code",
        "outputId": "dc25621b-1c83-479d-d4d1-bfe339ad5d1f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        }
      },
      "source": [
        "airline_tweets.airline.value_counts().plot(kind='pie', autopct='%1.1f%%')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7f435bfc1f28>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 52
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6gijP9lzoAWF",
        "colab_type": "code",
        "outputId": "637e628b-22b4-4875-dbda-52741e40b5c2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        }
      },
      "source": [
        "airline_tweets.airline_sentiment.value_counts().plot(kind='pie', colors= [\"red\", \"green\", \"yellow\"], autopct='%1.1f%%')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7f435bf5e048>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 53
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g5W-u3BCBfRN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Extracting features and label\n",
        "features = airline_tweets.iloc[:, 10].values\n",
        "labels = airline_tweets.iloc[:, 1].values"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oeDEmJ8yBiQj",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import re \n",
        "\n",
        "processed_features = []\n",
        "\n",
        "for sentence in range(0, len(features)):\n",
        "    # Remove all the special characters\n",
        "    processed_feature = re.sub(r'\\W', ' ', str(features[sentence]))\n",
        "\n",
        "    # remove all single characters\n",
        "    processed_feature= re.sub(r'\\s+[a-zA-Z]\\s+', ' ', processed_feature)\n",
        "\n",
        "    # Remove single characters from the start\n",
        "    processed_feature = re.sub(r'\\^[a-zA-Z]\\s+', ' ', processed_feature) \n",
        "\n",
        "    # Substituting multiple spaces with single space\n",
        "    processed_feature = re.sub(r'\\s+', ' ', processed_feature, flags=re.I)\n",
        "\n",
        "    # Removing prefixed 'b'\n",
        "    processed_feature = re.sub(r'^b\\s+', '', processed_feature)\n",
        "\n",
        "    # Converting to Lowercase\n",
        "    processed_feature = processed_feature.lower()\n",
        "\n",
        "    processed_features.append(processed_feature)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dft0VHwwujym",
        "colab_type": "code",
        "outputId": "52bef3b6-22d5-46fc-a53c-282d0cc3fd5c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "import nltk\n",
        "nltk.download('stopwords')\n",
        "from nltk.corpus import stopwords\n",
        "from sklearn.feature_extraction.text import TfidfVectorizer\n",
        "\n",
        "vectorizer= TfidfVectorizer(max_features=2500, min_df=7, max_df=0.8, \n",
        "        stop_words=stopwords.words('english'))\n",
        "processed_features= vectorizer.fit_transform(processed_features).toarray()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[nltk_data] Downloading package stopwords to /root/nltk_data...\n",
            "[nltk_data]   Package stopwords is already up-to-date!\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-a3bGRY35upV",
        "colab_type": "code",
        "outputId": "cccc351a-c984-41ba-f9ab-d29e319c2846",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 153
        }
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "X_train, X_test, y_train, y_test = train_test_split(processed_features, labels, \n",
        "                                                    test_size=0.2, random_state= 0)\n",
        "\n",
        "from sklearn.ensemble import RandomForestClassifier\n",
        "text_classifier = RandomForestClassifier(n_estimators=200, random_state=0)\n",
        "text_classifier.fit(X_train, y_train)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "RandomForestClassifier(bootstrap=True, ccp_alpha=0.0, class_weight=None,\n",
              "                       criterion='gini', max_depth=None, max_features='auto',\n",
              "                       max_leaf_nodes=None, max_samples=None,\n",
              "                       min_impurity_decrease=0.0, min_impurity_split=None,\n",
              "                       min_samples_leaf=1, min_samples_split=2,\n",
              "                       min_weight_fraction_leaf=0.0, n_estimators=200,\n",
              "                       n_jobs=None, oob_score=False, random_state=0, verbose=0,\n",
              "                       warm_start=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 85
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nEF46pxV6rVr",
        "colab_type": "code",
        "outputId": "2b51bab8-8534-4bb9-bd22-ce360397134d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 255
        }
      },
      "source": [
        "predictors= text_classifier.predict(X_test)\n",
        "\n",
        "from sklearn.metrics import classification_report, confusion_matrix, accuracy_score\n",
        "print(confusion_matrix(y_test, predictors))\n",
        "print(classification_report(y_test, predictors))\n",
        "print(accuracy_score(y_test, predictors))\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[1723  108   39]\n",
            " [ 326  248   40]\n",
            " [ 132   58  254]]\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "    negative       0.79      0.92      0.85      1870\n",
            "     neutral       0.60      0.40      0.48       614\n",
            "    positive       0.76      0.57      0.65       444\n",
            "\n",
            "    accuracy                           0.76      2928\n",
            "   macro avg       0.72      0.63      0.66      2928\n",
            "weighted avg       0.75      0.76      0.74      2928\n",
            "\n",
            "0.7599043715846995\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Awn9IT8q7jAF",
        "colab_type": "code",
        "outputId": "f6a39674-0e67-4c7c-cdd1-d25ab9f68479",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 153
        }
      },
      "source": [
        "text_classifier"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "RandomForestClassifier(bootstrap=True, ccp_alpha=0.0, class_weight=None,\n",
              "                       criterion='gini', max_depth=None, max_features='auto',\n",
              "                       max_leaf_nodes=None, max_samples=None,\n",
              "                       min_impurity_decrease=0.0, min_impurity_split=None,\n",
              "                       min_samples_leaf=1, min_samples_split=2,\n",
              "                       min_weight_fraction_leaf=0.0, n_estimators=200,\n",
              "                       n_jobs=None, oob_score=False, random_state=0, verbose=0,\n",
              "                       warm_start=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 87
        }
      ]
    }
  ]
}