Deep Learning Regression Dataset Auto MPG.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Assignment16_RickyGunawan_Auto_MPG.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ipe6byeT9FHw"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import os\n",
        "import seaborn as sb\n",
        "\n",
        "from sklearn import preprocessing\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.linear_model import LinearRegression\n",
        "from sklearn.metrics import mean_squared_error, mean_absolute_error, r2_score\n",
        "\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.layers import Dense\n",
        "from tensorflow.keras.optimizers import SGD\n",
        "from tensorflow.keras.optimizers import RMSprop"
      ],
      "execution_count": 281,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "niwiqZeU8Fby"
      },
      "source": [
        "# **Dataset Auto_MPG**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OaulOFjlpSGe",
        "outputId": "13aad2eb-fb1c-4b06-f159-54510fba6c13",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 266
        }
      },
      "source": [
        "dataset = pd.read_csv('/content/drive/My Drive/BCML/datasets/DeepLearning/auto-mpg.csv')\n",
        "\n",
        "dataset.describe().transpose()"
      ],
      "execution_count": 282,
      "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>count</th>\n",
              "      <th>mean</th>\n",
              "      <th>std</th>\n",
              "      <th>min</th>\n",
              "      <th>25%</th>\n",
              "      <th>50%</th>\n",
              "      <th>75%</th>\n",
              "      <th>max</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>cylinders</th>\n",
              "      <td>398.0</td>\n",
              "      <td>5.454774</td>\n",
              "      <td>1.701004</td>\n",
              "      <td>3.0</td>\n",
              "      <td>4.000</td>\n",
              "      <td>4.0</td>\n",
              "      <td>8.000</td>\n",
              "      <td>8.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>displacement</th>\n",
              "      <td>398.0</td>\n",
              "      <td>193.425879</td>\n",
              "      <td>104.269838</td>\n",
              "      <td>68.0</td>\n",
              "      <td>104.250</td>\n",
              "      <td>148.5</td>\n",
              "      <td>262.000</td>\n",
              "      <td>455.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>weight</th>\n",
              "      <td>398.0</td>\n",
              "      <td>2970.424623</td>\n",
              "      <td>846.841774</td>\n",
              "      <td>1613.0</td>\n",
              "      <td>2223.750</td>\n",
              "      <td>2803.5</td>\n",
              "      <td>3608.000</td>\n",
              "      <td>5140.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>acceleration</th>\n",
              "      <td>398.0</td>\n",
              "      <td>15.568090</td>\n",
              "      <td>2.757689</td>\n",
              "      <td>8.0</td>\n",
              "      <td>13.825</td>\n",
              "      <td>15.5</td>\n",
              "      <td>17.175</td>\n",
              "      <td>24.8</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>model year</th>\n",
              "      <td>398.0</td>\n",
              "      <td>76.010050</td>\n",
              "      <td>3.697627</td>\n",
              "      <td>70.0</td>\n",
              "      <td>73.000</td>\n",
              "      <td>76.0</td>\n",
              "      <td>79.000</td>\n",
              "      <td>82.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>origin</th>\n",
              "      <td>398.0</td>\n",
              "      <td>1.572864</td>\n",
              "      <td>0.802055</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.000</td>\n",
              "      <td>1.0</td>\n",
              "      <td>2.000</td>\n",
              "      <td>3.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mpg</th>\n",
              "      <td>398.0</td>\n",
              "      <td>23.514573</td>\n",
              "      <td>7.815984</td>\n",
              "      <td>9.0</td>\n",
              "      <td>17.500</td>\n",
              "      <td>23.0</td>\n",
              "      <td>29.000</td>\n",
              "      <td>46.6</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "              count         mean         std  ...     50%       75%     max\n",
              "cylinders     398.0     5.454774    1.701004  ...     4.0     8.000     8.0\n",
              "displacement  398.0   193.425879  104.269838  ...   148.5   262.000   455.0\n",
              "weight        398.0  2970.424623  846.841774  ...  2803.5  3608.000  5140.0\n",
              "acceleration  398.0    15.568090    2.757689  ...    15.5    17.175    24.8\n",
              "model year    398.0    76.010050    3.697627  ...    76.0    79.000    82.0\n",
              "origin        398.0     1.572864    0.802055  ...     1.0     2.000     3.0\n",
              "mpg           398.0    23.514573    7.815984  ...    23.0    29.000    46.6\n",
              "\n",
              "[7 rows x 8 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 282
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xRKLLi3w9msa",
        "outputId": "e74badfa-42e5-4409-aa39-591bf0695f73",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 289
        }
      },
      "source": [
        "dataset.info()"
      ],
      "execution_count": 283,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 398 entries, 0 to 397\n",
            "Data columns (total 9 columns):\n",
            " #   Column        Non-Null Count  Dtype  \n",
            "---  ------        --------------  -----  \n",
            " 0   cylinders     398 non-null    int64  \n",
            " 1   displacement  398 non-null    float64\n",
            " 2   horsepower    398 non-null    object \n",
            " 3   weight        398 non-null    int64  \n",
            " 4   acceleration  398 non-null    float64\n",
            " 5   model year    398 non-null    int64  \n",
            " 6   origin        398 non-null    int64  \n",
            " 7   car name      398 non-null    object \n",
            " 8   mpg           398 non-null    float64\n",
            "dtypes: float64(3), int64(4), object(2)\n",
            "memory usage: 28.1+ KB\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "T0Flg4Ik9owu",
        "outputId": "bebc5322-cc64-4699-ca9b-19566c2a7d3b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        }
      },
      "source": [
        "dataset.head()"
      ],
      "execution_count": 284,
      "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>cylinders</th>\n",
              "      <th>displacement</th>\n",
              "      <th>horsepower</th>\n",
              "      <th>weight</th>\n",
              "      <th>acceleration</th>\n",
              "      <th>model year</th>\n",
              "      <th>origin</th>\n",
              "      <th>car name</th>\n",
              "      <th>mpg</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>8</td>\n",
              "      <td>307.0</td>\n",
              "      <td>130</td>\n",
              "      <td>3504</td>\n",
              "      <td>12.0</td>\n",
              "      <td>70</td>\n",
              "      <td>1</td>\n",
              "      <td>chevrolet chevelle malibu</td>\n",
              "      <td>18.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>8</td>\n",
              "      <td>350.0</td>\n",
              "      <td>165</td>\n",
              "      <td>3693</td>\n",
              "      <td>11.5</td>\n",
              "      <td>70</td>\n",
              "      <td>1</td>\n",
              "      <td>buick skylark 320</td>\n",
              "      <td>15.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>8</td>\n",
              "      <td>318.0</td>\n",
              "      <td>150</td>\n",
              "      <td>3436</td>\n",
              "      <td>11.0</td>\n",
              "      <td>70</td>\n",
              "      <td>1</td>\n",
              "      <td>plymouth satellite</td>\n",
              "      <td>18.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>8</td>\n",
              "      <td>304.0</td>\n",
              "      <td>150</td>\n",
              "      <td>3433</td>\n",
              "      <td>12.0</td>\n",
              "      <td>70</td>\n",
              "      <td>1</td>\n",
              "      <td>amc rebel sst</td>\n",
              "      <td>16.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>8</td>\n",
              "      <td>302.0</td>\n",
              "      <td>140</td>\n",
              "      <td>3449</td>\n",
              "      <td>10.5</td>\n",
              "      <td>70</td>\n",
              "      <td>1</td>\n",
              "      <td>ford torino</td>\n",
              "      <td>17.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   cylinders  displacement horsepower  ...  origin                   car name   mpg\n",
              "0          8         307.0        130  ...       1  chevrolet chevelle malibu  18.0\n",
              "1          8         350.0        165  ...       1          buick skylark 320  15.0\n",
              "2          8         318.0        150  ...       1         plymouth satellite  18.0\n",
              "3          8         304.0        150  ...       1              amc rebel sst  16.0\n",
              "4          8         302.0        140  ...       1                ford torino  17.0\n",
              "\n",
              "[5 rows x 9 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 284
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yQlEnezC-y7S",
        "outputId": "5369c686-beb9-432c-a034-ec1b77d0af68",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        }
      },
      "source": [
        "dataset.replace('?',np.nan,inplace=True)\n",
        "dataset.isnull().sum().sort_values()"
      ],
      "execution_count": 285,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "cylinders       0\n",
              "displacement    0\n",
              "weight          0\n",
              "acceleration    0\n",
              "model year      0\n",
              "origin          0\n",
              "car name        0\n",
              "mpg             0\n",
              "horsepower      6\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 285
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jaoy-Tx7_J_0"
      },
      "source": [
        "for i in dataset.columns:\n",
        "    if dataset[i].isnull().sum() > 0:\n",
        "        dataset[i].fillna(dataset[i].astype(float).mean(), inplace=True)\n",
        "\n",
        "dataset[[\"horsepower\"]] = dataset[[\"horsepower\"]].astype(\"int\")\n"
      ],
      "execution_count": 286,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "depFb8a0_NN_",
        "outputId": "d15152b5-8672-46cf-e515-0487eb4f587c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        }
      },
      "source": [
        "dataset.isnull().sum().sort_values()"
      ],
      "execution_count": 287,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "cylinders       0\n",
              "displacement    0\n",
              "horsepower      0\n",
              "weight          0\n",
              "acceleration    0\n",
              "model year      0\n",
              "origin          0\n",
              "car name        0\n",
              "mpg             0\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 287
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EcliLtHjJ-sg",
        "outputId": "160b4ade-aed4-4046-bf69-f0fc4daf8b94",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 255
        }
      },
      "source": [
        "dataset['model year'].value_counts()"
      ],
      "execution_count": 288,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "73    40\n",
              "78    36\n",
              "76    34\n",
              "82    31\n",
              "75    30\n",
              "81    29\n",
              "80    29\n",
              "79    29\n",
              "70    29\n",
              "77    28\n",
              "72    28\n",
              "71    28\n",
              "74    27\n",
              "Name: model year, dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 288
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QHZ4qCyHLBM7"
      },
      "source": [
        "dataset.drop(['car name'],axis=1,inplace=True)"
      ],
      "execution_count": 289,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TAeT56yMLc7E",
        "outputId": "b17d7b69-b36c-4c6b-ba1f-e224580e576e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 272
        }
      },
      "source": [
        "dataset.info()"
      ],
      "execution_count": 290,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 398 entries, 0 to 397\n",
            "Data columns (total 8 columns):\n",
            " #   Column        Non-Null Count  Dtype  \n",
            "---  ------        --------------  -----  \n",
            " 0   cylinders     398 non-null    int64  \n",
            " 1   displacement  398 non-null    float64\n",
            " 2   horsepower    398 non-null    int64  \n",
            " 3   weight        398 non-null    int64  \n",
            " 4   acceleration  398 non-null    float64\n",
            " 5   model year    398 non-null    int64  \n",
            " 6   origin        398 non-null    int64  \n",
            " 7   mpg           398 non-null    float64\n",
            "dtypes: float64(3), int64(5)\n",
            "memory usage: 25.0 KB\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_Q2AyQHRLjWl",
        "outputId": "e5ddece0-32c0-4306-bf8d-a76bc9a822f8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "dataset.hist(figsize=(18, 22), bins=50, xlabelsize=8, ylabelsize=8)"
      ],
      "execution_count": 291,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7efbbdf80128>,\n",
              "        <matplotlib.axes._subplots.AxesSubplot object at 0x7efbe3a90eb8>,\n",
              "        <matplotlib.axes._subplots.AxesSubplot object at 0x7efbc794c278>],\n",
              "       [<matplotlib.axes._subplots.AxesSubplot object at 0x7efbe3ca05f8>,\n",
              "        <matplotlib.axes._subplots.AxesSubplot object at 0x7efbe3b78978>,\n",
              "        <matplotlib.axes._subplots.AxesSubplot object at 0x7efbe39fecf8>],\n",
              "       [<matplotlib.axes._subplots.AxesSubplot object at 0x7efbe3a220b8>,\n",
              "        <matplotlib.axes._subplots.AxesSubplot object at 0x7efbbd5583c8>,\n",
              "        <matplotlib.axes._subplots.AxesSubplot object at 0x7efbbd558438>]],\n",
              "      dtype=object)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 291
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1296x1584 with 9 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nAWdx_00Luxe",
        "outputId": "20928d97-a1fe-4797-f451-fdcd8ae1d9e0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "corr = dataset.corr()\n",
        "plt.figure(figsize=(20,20))\n",
        "sb.heatmap(corr, cbar=True, square= True, fmt='.1f', annot=True, annot_kws={'size':15}, cmap='Greens')"
      ],
      "execution_count": 292,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7efbbf9b51d0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 292
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x1440 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ap35nFgazKB5"
      },
      "source": [
        "# **Linear Regression**"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5Mzf-o6f0xVL"
      },
      "source": [
        "## **Feature Engineering**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uelGnt3DzImO"
      },
      "source": [
        "label = dataset['mpg']\n",
        "features = dataset.drop(['mpg'], axis=1)\n",
        "\n",
        "scaler = preprocessing.StandardScaler()\n",
        "\n",
        "features = scaler.fit_transform(features.values)\n",
        "label = scaler.fit_transform(label.values.reshape(-1,1)).flatten()\n",
        "\n",
        "feature_train, feature_test, label_train, label_test = train_test_split(features, label, test_size=0.2, random_state=10)"
      ],
      "execution_count": 293,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hlbqpuQu1emh",
        "outputId": "fa52618a-fa16-496e-e204-b20331d92f8f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        }
      },
      "source": [
        "linear_reg = LinearRegression()\n",
        "linear_reg.fit(feature_train, label_train)\n",
        "\n",
        "predictions = linear_reg.predict(feature_test)\n",
        "\n",
        "mse = mean_squared_error(label_test, predictions)\n",
        "mae = mean_absolute_error(label_test, predictions)\n",
        "r2 = r2_score(label_test, predictions)\n",
        "print(\"MSE (Mean Squared Error)\", mse)\n",
        "print(\"MAE (Mean Absolute Error)\", mae)\n",
        "print(\"r^2 score\", r2)\n",
        "print('RMSE (Root Mean Squared Error', np.sqrt(mean_squared_error(label_test, predictions)))"
      ],
      "execution_count": 294,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "MSE (Mean Squared Error) 0.20826660169611802\n",
            "MAE (Mean Absolute Error) 0.3684271684436153\n",
            "r^2 score 0.792478802889181\n",
            "RMSE (Root Mean Squared Error 0.4563623578869296\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JwwSzCjF5HCT"
      },
      "source": [
        "## **Base Model DL**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4SEGfnpB1z8b"
      },
      "source": [
        "model = Sequential()\n",
        "model.add(Dense(7,input_dim=7, kernel_initializer='normal', activation='relu')) # 13 neuron, sesuai dengan 13 features\n",
        "model.add(Dense(1, kernel_initializer='normal'))\n",
        "\n",
        "model.compile(loss='mean_squared_error', optimizer='adam')"
      ],
      "execution_count": 337,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gG_fsfis1_QE",
        "outputId": "8a03c993-29e9-4b56-c8a8-3144e7340f94",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "model.fit(x=feature_train, y=label_train, validation_data=(feature_test, label_test), epochs=50, batch_size=8)"
      ],
      "execution_count": 338,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "40/40 [==============================] - 0s 3ms/step - loss: 0.9671 - val_loss: 0.9335\n",
            "Epoch 2/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.8522 - val_loss: 0.7688\n",
            "Epoch 3/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.6309 - val_loss: 0.5088\n",
            "Epoch 4/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.3972 - val_loss: 0.3332\n",
            "Epoch 5/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.2799 - val_loss: 0.2736\n",
            "Epoch 6/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.2348 - val_loss: 0.2564\n",
            "Epoch 7/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.2180 - val_loss: 0.2457\n",
            "Epoch 8/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.2044 - val_loss: 0.2340\n",
            "Epoch 9/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1933 - val_loss: 0.2253\n",
            "Epoch 10/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1837 - val_loss: 0.2170\n",
            "Epoch 11/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1738 - val_loss: 0.2088\n",
            "Epoch 12/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1670 - val_loss: 0.2023\n",
            "Epoch 13/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1597 - val_loss: 0.1972\n",
            "Epoch 14/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1541 - val_loss: 0.1954\n",
            "Epoch 15/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1496 - val_loss: 0.1904\n",
            "Epoch 16/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1467 - val_loss: 0.1887\n",
            "Epoch 17/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1426 - val_loss: 0.1861\n",
            "Epoch 18/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1391 - val_loss: 0.1834\n",
            "Epoch 19/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1362 - val_loss: 0.1825\n",
            "Epoch 20/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1345 - val_loss: 0.1791\n",
            "Epoch 21/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1321 - val_loss: 0.1786\n",
            "Epoch 22/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1300 - val_loss: 0.1760\n",
            "Epoch 23/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1285 - val_loss: 0.1754\n",
            "Epoch 24/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1267 - val_loss: 0.1743\n",
            "Epoch 25/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1253 - val_loss: 0.1701\n",
            "Epoch 26/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1239 - val_loss: 0.1713\n",
            "Epoch 27/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1232 - val_loss: 0.1692\n",
            "Epoch 28/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1228 - val_loss: 0.1681\n",
            "Epoch 29/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1210 - val_loss: 0.1669\n",
            "Epoch 30/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1205 - val_loss: 0.1649\n",
            "Epoch 31/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1201 - val_loss: 0.1653\n",
            "Epoch 32/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1186 - val_loss: 0.1636\n",
            "Epoch 33/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1185 - val_loss: 0.1649\n",
            "Epoch 34/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1178 - val_loss: 0.1629\n",
            "Epoch 35/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1175 - val_loss: 0.1617\n",
            "Epoch 36/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1171 - val_loss: 0.1632\n",
            "Epoch 37/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1161 - val_loss: 0.1621\n",
            "Epoch 38/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1156 - val_loss: 0.1617\n",
            "Epoch 39/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1151 - val_loss: 0.1625\n",
            "Epoch 40/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1148 - val_loss: 0.1611\n",
            "Epoch 41/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1147 - val_loss: 0.1576\n",
            "Epoch 42/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1144 - val_loss: 0.1613\n",
            "Epoch 43/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1133 - val_loss: 0.1612\n",
            "Epoch 44/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1133 - val_loss: 0.1586\n",
            "Epoch 45/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1131 - val_loss: 0.1597\n",
            "Epoch 46/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1129 - val_loss: 0.1600\n",
            "Epoch 47/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1124 - val_loss: 0.1583\n",
            "Epoch 48/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1123 - val_loss: 0.1587\n",
            "Epoch 49/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1116 - val_loss: 0.1566\n",
            "Epoch 50/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1114 - val_loss: 0.1577\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7efbabf49080>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 338
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "obSwIi4G44Ig",
        "outputId": "c1af91e1-cc6d-4358-a736-62e3a6d7f29d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "predictions = model.predict(feature_test)\n",
        "predictions = predictions.flatten()\n",
        "\n",
        "mse = mean_squared_error(label_test, predictions)\n",
        "mae = mean_absolute_error(label_test, predictions)\n",
        "r2 = r2_score(label_test, predictions)\n",
        "print(\"MSE (Mean Squared Error)\", mse)\n",
        "print(\"MAE (Mean Absolute Error)\", mae)\n",
        "print(\"r^2 score\", r2)\n",
        "print('RMSE (Root Mean Squared Error', np.sqrt(mean_squared_error(label_test, predictions)))"
      ],
      "execution_count": 339,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:5 out of the last 13 calls to <function Model.make_predict_function.<locals>.predict_function at 0x7efbabfbb6a8> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/tutorials/customization/performance#python_or_tensor_args and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "MSE (Mean Squared Error) 0.15771143900367993\n",
            "MAE (Mean Absolute Error) 0.30847694243624807\n",
            "r^2 score 0.8428530241835525\n",
            "RMSE (Root Mean Squared Error 0.39712899542048036\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9mkPLcIj5MEA"
      },
      "source": [
        "## **Deeper Model DL**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8FDbHiHm5ZfE",
        "outputId": "cd27fb87-94ae-4360-ca3d-619d08db5171",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "deeper_model = Sequential()\n",
        "deeper_model.add(Dense(7,input_dim=7, kernel_initializer='normal', activation='relu'))\n",
        "deeper_model.add(Dense(6, kernel_initializer='normal', activation='relu'))\n",
        "deeper_model.add(Dense(1, kernel_initializer='normal'))\n",
        "\n",
        "deeper_model.compile(loss='mean_squared_error', optimizer='adam')\n",
        "\n",
        "deeper_model.fit(x=feature_train, y=label_train, validation_data=(feature_test, label_test), epochs=50, batch_size=8)"
      ],
      "execution_count": 340,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "40/40 [==============================] - 1s 19ms/step - loss: 0.9940 - val_loss: 0.9946\n",
            "Epoch 2/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.9602 - val_loss: 0.9308\n",
            "Epoch 3/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.8499 - val_loss: 0.7646\n",
            "Epoch 4/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.6466 - val_loss: 0.5412\n",
            "Epoch 5/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.4490 - val_loss: 0.3881\n",
            "Epoch 6/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.3343 - val_loss: 0.3179\n",
            "Epoch 7/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.2719 - val_loss: 0.2820\n",
            "Epoch 8/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.2446 - val_loss: 0.2680\n",
            "Epoch 9/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.2273 - val_loss: 0.2571\n",
            "Epoch 10/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.2148 - val_loss: 0.2447\n",
            "Epoch 11/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.2010 - val_loss: 0.2323\n",
            "Epoch 12/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1880 - val_loss: 0.2215\n",
            "Epoch 13/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1752 - val_loss: 0.2133\n",
            "Epoch 14/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1620 - val_loss: 0.2036\n",
            "Epoch 15/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1524 - val_loss: 0.1996\n",
            "Epoch 16/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1441 - val_loss: 0.1918\n",
            "Epoch 17/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1392 - val_loss: 0.1905\n",
            "Epoch 18/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1349 - val_loss: 0.1850\n",
            "Epoch 19/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1317 - val_loss: 0.1803\n",
            "Epoch 20/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1298 - val_loss: 0.1741\n",
            "Epoch 21/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1276 - val_loss: 0.1679\n",
            "Epoch 22/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1246 - val_loss: 0.1650\n",
            "Epoch 23/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1231 - val_loss: 0.1630\n",
            "Epoch 24/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1208 - val_loss: 0.1622\n",
            "Epoch 25/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1186 - val_loss: 0.1575\n",
            "Epoch 26/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1164 - val_loss: 0.1588\n",
            "Epoch 27/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1150 - val_loss: 0.1582\n",
            "Epoch 28/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1141 - val_loss: 0.1587\n",
            "Epoch 29/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1140 - val_loss: 0.1586\n",
            "Epoch 30/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1128 - val_loss: 0.1568\n",
            "Epoch 31/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1121 - val_loss: 0.1566\n",
            "Epoch 32/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1109 - val_loss: 0.1547\n",
            "Epoch 33/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1114 - val_loss: 0.1559\n",
            "Epoch 34/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1102 - val_loss: 0.1529\n",
            "Epoch 35/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1100 - val_loss: 0.1535\n",
            "Epoch 36/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1091 - val_loss: 0.1567\n",
            "Epoch 37/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1090 - val_loss: 0.1537\n",
            "Epoch 38/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1080 - val_loss: 0.1578\n",
            "Epoch 39/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1076 - val_loss: 0.1547\n",
            "Epoch 40/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1080 - val_loss: 0.1583\n",
            "Epoch 41/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1067 - val_loss: 0.1545\n",
            "Epoch 42/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1065 - val_loss: 0.1569\n",
            "Epoch 43/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1060 - val_loss: 0.1539\n",
            "Epoch 44/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1047 - val_loss: 0.1551\n",
            "Epoch 45/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1064 - val_loss: 0.1554\n",
            "Epoch 46/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1040 - val_loss: 0.1533\n",
            "Epoch 47/50\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1030 - val_loss: 0.1569\n",
            "Epoch 48/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1034 - val_loss: 0.1556\n",
            "Epoch 49/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1025 - val_loss: 0.1552\n",
            "Epoch 50/50\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1021 - val_loss: 0.1577\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7efbab65e518>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 340
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h86IPhQ25iix",
        "outputId": "baee161e-894e-4142-820e-416d2fc109c0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "predictions = deeper_model.predict(feature_test)\n",
        "predictions = predictions.flatten()\n",
        "\n",
        "mse = mean_squared_error(label_test, predictions)\n",
        "mae = mean_absolute_error(label_test, predictions)\n",
        "r2 = r2_score(label_test, predictions)\n",
        "print(\"MSE (Mean Squared Error)\", mse)\n",
        "print(\"MAE (Mean Absolute Error)\", mae)\n",
        "print(\"r^2 score\", r2)\n",
        "print('RMSE (Root Mean Squared Error', np.sqrt(mean_squared_error(label_test, predictions)))"
      ],
      "execution_count": 341,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:5 out of the last 13 calls to <function Model.make_predict_function.<locals>.predict_function at 0x7efbabf7e8c8> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/tutorials/customization/performance#python_or_tensor_args and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "MSE (Mean Squared Error) 0.1576851553540551\n",
            "MAE (Mean Absolute Error) 0.3070258370985159\n",
            "r^2 score 0.842879213761671\n",
            "RMSE (Root Mean Squared Error 0.39709590196079225\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mhI5GNRM5n9Q"
      },
      "source": [
        "## **Wider Model DL**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hfxp5yDv5nM0",
        "outputId": "8225a74e-b8bf-486d-9f7b-153e9b31df86",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "wider_model = Sequential()\n",
        "wider_model.add(Dense(20,input_dim=7, kernel_initializer='normal', activation='relu'))\n",
        "wider_model.add(Dense(1, kernel_initializer='normal'))\n",
        "\n",
        "wider_model.compile(loss='mean_squared_error', optimizer='adam')\n",
        "\n",
        "history = wider_model.fit(x=feature_train, y=label_train, validation_data=(feature_test, label_test), epochs=100, batch_size=8)"
      ],
      "execution_count": 342,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "40/40 [==============================] - 0s 4ms/step - loss: 0.8721 - val_loss: 0.7654\n",
            "Epoch 2/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.5862 - val_loss: 0.4404\n",
            "Epoch 3/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.3141 - val_loss: 0.2822\n",
            "Epoch 4/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.2335 - val_loss: 0.2528\n",
            "Epoch 5/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.2079 - val_loss: 0.2327\n",
            "Epoch 6/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1888 - val_loss: 0.2137\n",
            "Epoch 7/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1733 - val_loss: 0.2004\n",
            "Epoch 8/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1592 - val_loss: 0.1902\n",
            "Epoch 9/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1478 - val_loss: 0.1827\n",
            "Epoch 10/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1390 - val_loss: 0.1764\n",
            "Epoch 11/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1343 - val_loss: 0.1779\n",
            "Epoch 12/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1301 - val_loss: 0.1702\n",
            "Epoch 13/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1263 - val_loss: 0.1670\n",
            "Epoch 14/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1238 - val_loss: 0.1630\n",
            "Epoch 15/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1218 - val_loss: 0.1639\n",
            "Epoch 16/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1205 - val_loss: 0.1671\n",
            "Epoch 17/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1202 - val_loss: 0.1638\n",
            "Epoch 18/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1162 - val_loss: 0.1618\n",
            "Epoch 19/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1154 - val_loss: 0.1610\n",
            "Epoch 20/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1135 - val_loss: 0.1603\n",
            "Epoch 21/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1129 - val_loss: 0.1563\n",
            "Epoch 22/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1114 - val_loss: 0.1595\n",
            "Epoch 23/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1110 - val_loss: 0.1558\n",
            "Epoch 24/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1108 - val_loss: 0.1566\n",
            "Epoch 25/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1107 - val_loss: 0.1560\n",
            "Epoch 26/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1088 - val_loss: 0.1596\n",
            "Epoch 27/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1082 - val_loss: 0.1533\n",
            "Epoch 28/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1083 - val_loss: 0.1529\n",
            "Epoch 29/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1070 - val_loss: 0.1541\n",
            "Epoch 30/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1063 - val_loss: 0.1580\n",
            "Epoch 31/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1065 - val_loss: 0.1534\n",
            "Epoch 32/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1070 - val_loss: 0.1604\n",
            "Epoch 33/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1058 - val_loss: 0.1506\n",
            "Epoch 34/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1051 - val_loss: 0.1544\n",
            "Epoch 35/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1052 - val_loss: 0.1556\n",
            "Epoch 36/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1045 - val_loss: 0.1553\n",
            "Epoch 37/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1052 - val_loss: 0.1541\n",
            "Epoch 38/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1038 - val_loss: 0.1533\n",
            "Epoch 39/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1041 - val_loss: 0.1527\n",
            "Epoch 40/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1034 - val_loss: 0.1496\n",
            "Epoch 41/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1033 - val_loss: 0.1545\n",
            "Epoch 42/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1027 - val_loss: 0.1549\n",
            "Epoch 43/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1019 - val_loss: 0.1538\n",
            "Epoch 44/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1038 - val_loss: 0.1571\n",
            "Epoch 45/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1018 - val_loss: 0.1523\n",
            "Epoch 46/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1020 - val_loss: 0.1519\n",
            "Epoch 47/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1015 - val_loss: 0.1520\n",
            "Epoch 48/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1026 - val_loss: 0.1482\n",
            "Epoch 49/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1022 - val_loss: 0.1506\n",
            "Epoch 50/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1013 - val_loss: 0.1539\n",
            "Epoch 51/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1015 - val_loss: 0.1533\n",
            "Epoch 52/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1014 - val_loss: 0.1514\n",
            "Epoch 53/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1010 - val_loss: 0.1572\n",
            "Epoch 54/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1023 - val_loss: 0.1522\n",
            "Epoch 55/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1009 - val_loss: 0.1492\n",
            "Epoch 56/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1024 - val_loss: 0.1488\n",
            "Epoch 57/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1017 - val_loss: 0.1511\n",
            "Epoch 58/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0994 - val_loss: 0.1515\n",
            "Epoch 59/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0997 - val_loss: 0.1542\n",
            "Epoch 60/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1000 - val_loss: 0.1480\n",
            "Epoch 61/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0992 - val_loss: 0.1518\n",
            "Epoch 62/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0994 - val_loss: 0.1551\n",
            "Epoch 63/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1001 - val_loss: 0.1570\n",
            "Epoch 64/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0985 - val_loss: 0.1497\n",
            "Epoch 65/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0990 - val_loss: 0.1541\n",
            "Epoch 66/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0977 - val_loss: 0.1507\n",
            "Epoch 67/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0985 - val_loss: 0.1542\n",
            "Epoch 68/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0973 - val_loss: 0.1506\n",
            "Epoch 69/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0978 - val_loss: 0.1535\n",
            "Epoch 70/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0984 - val_loss: 0.1607\n",
            "Epoch 71/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0998 - val_loss: 0.1591\n",
            "Epoch 72/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0977 - val_loss: 0.1535\n",
            "Epoch 73/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0979 - val_loss: 0.1507\n",
            "Epoch 74/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0977 - val_loss: 0.1569\n",
            "Epoch 75/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0969 - val_loss: 0.1503\n",
            "Epoch 76/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0967 - val_loss: 0.1492\n",
            "Epoch 77/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0983 - val_loss: 0.1456\n",
            "Epoch 78/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0964 - val_loss: 0.1513\n",
            "Epoch 79/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0963 - val_loss: 0.1488\n",
            "Epoch 80/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0957 - val_loss: 0.1554\n",
            "Epoch 81/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0967 - val_loss: 0.1467\n",
            "Epoch 82/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0950 - val_loss: 0.1508\n",
            "Epoch 83/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0961 - val_loss: 0.1522\n",
            "Epoch 84/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0958 - val_loss: 0.1534\n",
            "Epoch 85/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0948 - val_loss: 0.1469\n",
            "Epoch 86/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0950 - val_loss: 0.1596\n",
            "Epoch 87/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0969 - val_loss: 0.1486\n",
            "Epoch 88/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0948 - val_loss: 0.1490\n",
            "Epoch 89/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0945 - val_loss: 0.1556\n",
            "Epoch 90/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0946 - val_loss: 0.1449\n",
            "Epoch 91/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0940 - val_loss: 0.1494\n",
            "Epoch 92/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0961 - val_loss: 0.1514\n",
            "Epoch 93/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0947 - val_loss: 0.1500\n",
            "Epoch 94/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0946 - val_loss: 0.1492\n",
            "Epoch 95/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0938 - val_loss: 0.1470\n",
            "Epoch 96/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0953 - val_loss: 0.1464\n",
            "Epoch 97/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0947 - val_loss: 0.1540\n",
            "Epoch 98/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0931 - val_loss: 0.1489\n",
            "Epoch 99/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0934 - val_loss: 0.1503\n",
            "Epoch 100/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0931 - val_loss: 0.1485\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tgkMPhT063sq",
        "outputId": "ba54b674-6d81-4531-c25c-5d526298b6a4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "predictions = wider_model.predict(feature_test)\n",
        "predictions = predictions.flatten()\n",
        "\n",
        "mse = mean_squared_error(label_test, predictions)\n",
        "mae = mean_absolute_error(label_test, predictions)\n",
        "r2 = r2_score(label_test, predictions)\n",
        "print(\"MSE (Mean Squared Error)\", mse)\n",
        "print(\"MAE (Mean Absolute Error)\", mae)\n",
        "print(\"r^2 score\", r2)\n",
        "print('RMSE (Root Mean Squared Error', np.sqrt(mean_squared_error(label_test, predictions)))"
      ],
      "execution_count": 343,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:5 out of the last 13 calls to <function Model.make_predict_function.<locals>.predict_function at 0x7efbae9ab510> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/tutorials/customization/performance#python_or_tensor_args and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "MSE (Mean Squared Error) 0.14851992723902346\n",
            "MAE (Mean Absolute Error) 0.2919318463297486\n",
            "r^2 score 0.8520116387147619\n",
            "RMSE (Root Mean Squared Error 0.3853828320501881\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yUB62LGt7Pz6"
      },
      "source": [
        "def plot_loss(history):\n",
        "  plt.plot(history.history['loss'], label='loss')\n",
        "  plt.plot(history.history['val_loss'], label='val_loss')\n",
        "  plt.xlabel('Epoch')\n",
        "  plt.ylabel('Loss (MSE)')\n",
        "  plt.legend()\n",
        "  plt.grid(True)"
      ],
      "execution_count": 344,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vkPCFAle7Q_C",
        "outputId": "3ab20a62-69fd-400c-e26a-3405cd644989",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        }
      },
      "source": [
        "history_dataframe = pd.DataFrame(history.history)\n",
        "history_dataframe['epoch'] = history.epoch\n",
        "history_dataframe"
      ],
      "execution_count": 345,
      "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>loss</th>\n",
              "      <th>val_loss</th>\n",
              "      <th>epoch</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.872061</td>\n",
              "      <td>0.765351</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.586248</td>\n",
              "      <td>0.440375</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.314134</td>\n",
              "      <td>0.282195</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.233519</td>\n",
              "      <td>0.252825</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0.207878</td>\n",
              "      <td>0.232710</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>95</th>\n",
              "      <td>0.095252</td>\n",
              "      <td>0.146405</td>\n",
              "      <td>95</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>96</th>\n",
              "      <td>0.094707</td>\n",
              "      <td>0.154047</td>\n",
              "      <td>96</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>97</th>\n",
              "      <td>0.093074</td>\n",
              "      <td>0.148913</td>\n",
              "      <td>97</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>98</th>\n",
              "      <td>0.093423</td>\n",
              "      <td>0.150290</td>\n",
              "      <td>98</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>99</th>\n",
              "      <td>0.093061</td>\n",
              "      <td>0.148520</td>\n",
              "      <td>99</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>100 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "        loss  val_loss  epoch\n",
              "0   0.872061  0.765351      0\n",
              "1   0.586248  0.440375      1\n",
              "2   0.314134  0.282195      2\n",
              "3   0.233519  0.252825      3\n",
              "4   0.207878  0.232710      4\n",
              "..       ...       ...    ...\n",
              "95  0.095252  0.146405     95\n",
              "96  0.094707  0.154047     96\n",
              "97  0.093074  0.148913     97\n",
              "98  0.093423  0.150290     98\n",
              "99  0.093061  0.148520     99\n",
              "\n",
              "[100 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 345
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2Rupgn9opHD3",
        "outputId": "f4d8563c-0cd5-4525-893a-83a02e3851bc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        }
      },
      "source": [
        "history_dataframe.sort_values(by='val_loss', ascending=True)"
      ],
      "execution_count": 346,
      "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>loss</th>\n",
              "      <th>val_loss</th>\n",
              "      <th>epoch</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>89</th>\n",
              "      <td>0.094586</td>\n",
              "      <td>0.144865</td>\n",
              "      <td>89</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>76</th>\n",
              "      <td>0.098287</td>\n",
              "      <td>0.145587</td>\n",
              "      <td>76</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>95</th>\n",
              "      <td>0.095252</td>\n",
              "      <td>0.146405</td>\n",
              "      <td>95</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>80</th>\n",
              "      <td>0.096699</td>\n",
              "      <td>0.146677</td>\n",
              "      <td>80</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>84</th>\n",
              "      <td>0.094827</td>\n",
              "      <td>0.146909</td>\n",
              "      <td>84</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0.207878</td>\n",
              "      <td>0.232710</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.233519</td>\n",
              "      <td>0.252825</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.314134</td>\n",
              "      <td>0.282195</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.586248</td>\n",
              "      <td>0.440375</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.872061</td>\n",
              "      <td>0.765351</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>100 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "        loss  val_loss  epoch\n",
              "89  0.094586  0.144865     89\n",
              "76  0.098287  0.145587     76\n",
              "95  0.095252  0.146405     95\n",
              "80  0.096699  0.146677     80\n",
              "84  0.094827  0.146909     84\n",
              "..       ...       ...    ...\n",
              "4   0.207878  0.232710      4\n",
              "3   0.233519  0.252825      3\n",
              "2   0.314134  0.282195      2\n",
              "1   0.586248  0.440375      1\n",
              "0   0.872061  0.765351      0\n",
              "\n",
              "[100 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 346
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_2rnNXWY7Wmq",
        "outputId": "d2c3ba6a-ec0a-4657-9bc5-95b6645290d2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        }
      },
      "source": [
        "plot_loss(history) # epoch vs loss graph"
      ],
      "execution_count": 347,
      "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": "markdown",
      "metadata": {
        "id": "7TDa0NNzj2H3"
      },
      "source": [
        "### **Optimizer SGD**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LJS7Hqjbj1QI",
        "outputId": "4bcc08b7-19af-4bb5-a507-eebcb193d9f5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 221
        }
      },
      "source": [
        "wider_model_tuned = Sequential()\n",
        "wider_model_tuned.add(Dense(20,input_dim=7, kernel_initializer='normal', activation='relu'))\n",
        "wider_model_tuned.add(Dense(1, kernel_initializer='normal'))\n",
        "\n",
        "opt = SGD(lr=0.001, momentum=0.9)\n",
        "\n",
        "wider_model_tuned.compile(loss='mean_squared_error', optimizer=opt)\n",
        "\n",
        "wider_model_tuned.summary()"
      ],
      "execution_count": 348,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_53\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "dense_117 (Dense)            (None, 20)                160       \n",
            "_________________________________________________________________\n",
            "dense_118 (Dense)            (None, 1)                 21        \n",
            "=================================================================\n",
            "Total params: 181\n",
            "Trainable params: 181\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TEbFbfYAkF2d",
        "outputId": "cae6b288-18a7-4c42-8a33-e89dfe1d3a2e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "history_tuned = wider_model_tuned.fit(x=feature_train, y=label_train, validation_data=(feature_test, label_test), epochs=100, batch_size=8)"
      ],
      "execution_count": 349,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "40/40 [==============================] - 0s 3ms/step - loss: 0.9430 - val_loss: 0.8600\n",
            "Epoch 2/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.7297 - val_loss: 0.5737\n",
            "Epoch 3/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.4391 - val_loss: 0.3365\n",
            "Epoch 4/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.2775 - val_loss: 0.2754\n",
            "Epoch 5/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.2311 - val_loss: 0.2519\n",
            "Epoch 6/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.2083 - val_loss: 0.2296\n",
            "Epoch 7/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1886 - val_loss: 0.2147\n",
            "Epoch 8/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1754 - val_loss: 0.2038\n",
            "Epoch 9/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1647 - val_loss: 0.1969\n",
            "Epoch 10/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1566 - val_loss: 0.1876\n",
            "Epoch 11/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1516 - val_loss: 0.1830\n",
            "Epoch 12/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1464 - val_loss: 0.1779\n",
            "Epoch 13/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1439 - val_loss: 0.1737\n",
            "Epoch 14/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1398 - val_loss: 0.1736\n",
            "Epoch 15/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1384 - val_loss: 0.1725\n",
            "Epoch 16/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1357 - val_loss: 0.1706\n",
            "Epoch 17/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1339 - val_loss: 0.1647\n",
            "Epoch 18/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1327 - val_loss: 0.1652\n",
            "Epoch 19/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1310 - val_loss: 0.1647\n",
            "Epoch 20/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1292 - val_loss: 0.1654\n",
            "Epoch 21/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1279 - val_loss: 0.1620\n",
            "Epoch 22/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1267 - val_loss: 0.1626\n",
            "Epoch 23/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1266 - val_loss: 0.1657\n",
            "Epoch 24/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1245 - val_loss: 0.1598\n",
            "Epoch 25/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1234 - val_loss: 0.1606\n",
            "Epoch 26/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1224 - val_loss: 0.1605\n",
            "Epoch 27/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1237 - val_loss: 0.1571\n",
            "Epoch 28/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1208 - val_loss: 0.1616\n",
            "Epoch 29/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1205 - val_loss: 0.1567\n",
            "Epoch 30/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1194 - val_loss: 0.1594\n",
            "Epoch 31/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1183 - val_loss: 0.1562\n",
            "Epoch 32/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1179 - val_loss: 0.1579\n",
            "Epoch 33/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1171 - val_loss: 0.1586\n",
            "Epoch 34/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1167 - val_loss: 0.1580\n",
            "Epoch 35/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1157 - val_loss: 0.1595\n",
            "Epoch 36/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1156 - val_loss: 0.1570\n",
            "Epoch 37/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1154 - val_loss: 0.1545\n",
            "Epoch 38/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1150 - val_loss: 0.1532\n",
            "Epoch 39/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1138 - val_loss: 0.1614\n",
            "Epoch 40/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1139 - val_loss: 0.1594\n",
            "Epoch 41/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1131 - val_loss: 0.1546\n",
            "Epoch 42/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1127 - val_loss: 0.1576\n",
            "Epoch 43/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1117 - val_loss: 0.1591\n",
            "Epoch 44/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1114 - val_loss: 0.1561\n",
            "Epoch 45/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1110 - val_loss: 0.1551\n",
            "Epoch 46/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1110 - val_loss: 0.1574\n",
            "Epoch 47/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1104 - val_loss: 0.1573\n",
            "Epoch 48/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1108 - val_loss: 0.1554\n",
            "Epoch 49/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1105 - val_loss: 0.1554\n",
            "Epoch 50/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1097 - val_loss: 0.1580\n",
            "Epoch 51/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1094 - val_loss: 0.1534\n",
            "Epoch 52/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1096 - val_loss: 0.1562\n",
            "Epoch 53/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1091 - val_loss: 0.1551\n",
            "Epoch 54/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1093 - val_loss: 0.1554\n",
            "Epoch 55/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1088 - val_loss: 0.1533\n",
            "Epoch 56/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1083 - val_loss: 0.1586\n",
            "Epoch 57/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1095 - val_loss: 0.1524\n",
            "Epoch 58/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1075 - val_loss: 0.1603\n",
            "Epoch 59/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1088 - val_loss: 0.1569\n",
            "Epoch 60/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1079 - val_loss: 0.1569\n",
            "Epoch 61/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1064 - val_loss: 0.1521\n",
            "Epoch 62/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1079 - val_loss: 0.1549\n",
            "Epoch 63/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1096 - val_loss: 0.1583\n",
            "Epoch 64/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1063 - val_loss: 0.1543\n",
            "Epoch 65/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1074 - val_loss: 0.1524\n",
            "Epoch 66/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1059 - val_loss: 0.1540\n",
            "Epoch 67/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1062 - val_loss: 0.1516\n",
            "Epoch 68/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1056 - val_loss: 0.1540\n",
            "Epoch 69/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1056 - val_loss: 0.1526\n",
            "Epoch 70/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1057 - val_loss: 0.1551\n",
            "Epoch 71/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1052 - val_loss: 0.1559\n",
            "Epoch 72/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1059 - val_loss: 0.1569\n",
            "Epoch 73/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1042 - val_loss: 0.1521\n",
            "Epoch 74/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1050 - val_loss: 0.1531\n",
            "Epoch 75/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1047 - val_loss: 0.1523\n",
            "Epoch 76/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1046 - val_loss: 0.1540\n",
            "Epoch 77/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1051 - val_loss: 0.1545\n",
            "Epoch 78/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1050 - val_loss: 0.1531\n",
            "Epoch 79/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1043 - val_loss: 0.1525\n",
            "Epoch 80/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1047 - val_loss: 0.1560\n",
            "Epoch 81/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1040 - val_loss: 0.1517\n",
            "Epoch 82/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1039 - val_loss: 0.1562\n",
            "Epoch 83/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1034 - val_loss: 0.1551\n",
            "Epoch 84/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1037 - val_loss: 0.1531\n",
            "Epoch 85/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1034 - val_loss: 0.1530\n",
            "Epoch 86/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1030 - val_loss: 0.1557\n",
            "Epoch 87/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1028 - val_loss: 0.1527\n",
            "Epoch 88/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1026 - val_loss: 0.1533\n",
            "Epoch 89/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1028 - val_loss: 0.1535\n",
            "Epoch 90/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1027 - val_loss: 0.1519\n",
            "Epoch 91/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1026 - val_loss: 0.1506\n",
            "Epoch 92/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1021 - val_loss: 0.1519\n",
            "Epoch 93/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1027 - val_loss: 0.1520\n",
            "Epoch 94/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1031 - val_loss: 0.1488\n",
            "Epoch 95/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1030 - val_loss: 0.1558\n",
            "Epoch 96/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1021 - val_loss: 0.1510\n",
            "Epoch 97/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1010 - val_loss: 0.1549\n",
            "Epoch 98/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1020 - val_loss: 0.1515\n",
            "Epoch 99/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1013 - val_loss: 0.1518\n",
            "Epoch 100/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1014 - val_loss: 0.1509\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "y_GR4CHxlIdR",
        "outputId": "9f6e36cb-c311-4dd0-91a3-b7a40d80c234",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "predictions = wider_model_tuned.predict(feature_test)\n",
        "predictions = predictions.flatten()\n",
        "\n",
        "mse = mean_squared_error(label_test, predictions)\n",
        "mae = mean_absolute_error(label_test, predictions)\n",
        "r2 = r2_score(label_test, predictions)\n",
        "print(\"MSE (Mean Squared Error)\", mse)\n",
        "print(\"MAE (Mean Absolute Error)\", mae)\n",
        "print(\"r^2 score\", r2)\n",
        "print('RMSE (Root Mean Squared Error', np.sqrt(mean_squared_error(label_test, predictions)))"
      ],
      "execution_count": 350,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:5 out of the last 13 calls to <function Model.make_predict_function.<locals>.predict_function at 0x7efbbbf2bd90> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/tutorials/customization/performance#python_or_tensor_args and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "MSE (Mean Squared Error) 0.15093339263924638\n",
            "MAE (Mean Absolute Error) 0.29637474855415136\n",
            "r^2 score 0.8496068113206386\n",
            "RMSE (Root Mean Squared Error 0.3885014705753974\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "a4q6XNHqkQXF",
        "outputId": "240dd240-b08e-41cd-b3e1-f65201388ba7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        }
      },
      "source": [
        "history_tuned_dataframe = pd.DataFrame(history_tuned.history)\n",
        "history_tuned_dataframe['epoch'] = history_tuned.epoch\n",
        "history_tuned_dataframe.sort_values(by='val_loss', ascending=True)"
      ],
      "execution_count": 351,
      "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>loss</th>\n",
              "      <th>val_loss</th>\n",
              "      <th>epoch</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>93</th>\n",
              "      <td>0.103142</td>\n",
              "      <td>0.148815</td>\n",
              "      <td>93</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>90</th>\n",
              "      <td>0.102626</td>\n",
              "      <td>0.150622</td>\n",
              "      <td>90</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>99</th>\n",
              "      <td>0.101392</td>\n",
              "      <td>0.150933</td>\n",
              "      <td>99</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>95</th>\n",
              "      <td>0.102147</td>\n",
              "      <td>0.150969</td>\n",
              "      <td>95</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>97</th>\n",
              "      <td>0.102019</td>\n",
              "      <td>0.151527</td>\n",
              "      <td>97</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0.231142</td>\n",
              "      <td>0.251853</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.277507</td>\n",
              "      <td>0.275440</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.439147</td>\n",
              "      <td>0.336478</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.729658</td>\n",
              "      <td>0.573728</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.942984</td>\n",
              "      <td>0.860041</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>100 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "        loss  val_loss  epoch\n",
              "93  0.103142  0.148815     93\n",
              "90  0.102626  0.150622     90\n",
              "99  0.101392  0.150933     99\n",
              "95  0.102147  0.150969     95\n",
              "97  0.102019  0.151527     97\n",
              "..       ...       ...    ...\n",
              "4   0.231142  0.251853      4\n",
              "3   0.277507  0.275440      3\n",
              "2   0.439147  0.336478      2\n",
              "1   0.729658  0.573728      1\n",
              "0   0.942984  0.860041      0\n",
              "\n",
              "[100 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 351
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fB0Qkh28kafu",
        "outputId": "4bfd90bc-48b8-42b8-d904-be9c08ba3e6e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        }
      },
      "source": [
        "plot_loss(history_tuned) # epoch vs loss graph"
      ],
      "execution_count": 352,
      "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": "markdown",
      "metadata": {
        "id": "-CeJOhqpktXe"
      },
      "source": [
        "### **Optimizer RMSprop**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gs4YeZEEk18f",
        "outputId": "5df45f58-64de-41af-c7d1-3636f3aa703a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 221
        }
      },
      "source": [
        "wider_model_rmsprop = Sequential()\n",
        "wider_model_rmsprop.add(Dense(20, input_dim=7, kernel_initializer='normal', activation='relu'))\n",
        "wider_model_rmsprop.add(Dense(1, kernel_initializer='normal'))\n",
        "\n",
        "opt = RMSprop(lr=0.001)\n",
        "\n",
        "wider_model_rmsprop.compile(loss='mean_squared_error', optimizer=opt)\n",
        "\n",
        "wider_model_rmsprop.summary()"
      ],
      "execution_count": 353,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_54\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "dense_119 (Dense)            (None, 20)                160       \n",
            "_________________________________________________________________\n",
            "dense_120 (Dense)            (None, 1)                 21        \n",
            "=================================================================\n",
            "Total params: 181\n",
            "Trainable params: 181\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2Wao2ul6k6yd",
        "outputId": "01589b30-78a7-438c-ca07-3f10676f241f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "history_rmsprop = wider_model_rmsprop.fit(x=feature_train, y=label_train, validation_data=(feature_test, label_test), epochs=100, batch_size=8)"
      ],
      "execution_count": 354,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "40/40 [==============================] - 0s 4ms/step - loss: 0.8429 - val_loss: 0.6935\n",
            "Epoch 2/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.5259 - val_loss: 0.3928\n",
            "Epoch 3/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.2911 - val_loss: 0.2655\n",
            "Epoch 4/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.2225 - val_loss: 0.2406\n",
            "Epoch 5/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1983 - val_loss: 0.2209\n",
            "Epoch 6/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1785 - val_loss: 0.2067\n",
            "Epoch 7/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1639 - val_loss: 0.1968\n",
            "Epoch 8/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1533 - val_loss: 0.1887\n",
            "Epoch 9/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1457 - val_loss: 0.1803\n",
            "Epoch 10/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1394 - val_loss: 0.1759\n",
            "Epoch 11/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1354 - val_loss: 0.1731\n",
            "Epoch 12/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1315 - val_loss: 0.1691\n",
            "Epoch 13/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1298 - val_loss: 0.1736\n",
            "Epoch 14/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1270 - val_loss: 0.1681\n",
            "Epoch 15/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1259 - val_loss: 0.1661\n",
            "Epoch 16/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1235 - val_loss: 0.1646\n",
            "Epoch 17/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1217 - val_loss: 0.1615\n",
            "Epoch 18/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1200 - val_loss: 0.1680\n",
            "Epoch 19/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1185 - val_loss: 0.1669\n",
            "Epoch 20/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1177 - val_loss: 0.1590\n",
            "Epoch 21/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1167 - val_loss: 0.1607\n",
            "Epoch 22/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1159 - val_loss: 0.1614\n",
            "Epoch 23/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1143 - val_loss: 0.1664\n",
            "Epoch 24/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1140 - val_loss: 0.1596\n",
            "Epoch 25/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1134 - val_loss: 0.1572\n",
            "Epoch 26/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1128 - val_loss: 0.1607\n",
            "Epoch 27/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1129 - val_loss: 0.1572\n",
            "Epoch 28/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1114 - val_loss: 0.1558\n",
            "Epoch 29/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1101 - val_loss: 0.1606\n",
            "Epoch 30/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1099 - val_loss: 0.1555\n",
            "Epoch 31/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1100 - val_loss: 0.1600\n",
            "Epoch 32/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1095 - val_loss: 0.1555\n",
            "Epoch 33/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1084 - val_loss: 0.1522\n",
            "Epoch 34/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1078 - val_loss: 0.1607\n",
            "Epoch 35/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1072 - val_loss: 0.1551\n",
            "Epoch 36/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1064 - val_loss: 0.1499\n",
            "Epoch 37/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1068 - val_loss: 0.1542\n",
            "Epoch 38/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1053 - val_loss: 0.1486\n",
            "Epoch 39/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1058 - val_loss: 0.1519\n",
            "Epoch 40/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1047 - val_loss: 0.1557\n",
            "Epoch 41/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1054 - val_loss: 0.1516\n",
            "Epoch 42/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1041 - val_loss: 0.1514\n",
            "Epoch 43/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1041 - val_loss: 0.1513\n",
            "Epoch 44/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1039 - val_loss: 0.1539\n",
            "Epoch 45/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1028 - val_loss: 0.1538\n",
            "Epoch 46/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1034 - val_loss: 0.1537\n",
            "Epoch 47/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1029 - val_loss: 0.1506\n",
            "Epoch 48/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1035 - val_loss: 0.1515\n",
            "Epoch 49/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1007 - val_loss: 0.1544\n",
            "Epoch 50/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1019 - val_loss: 0.1536\n",
            "Epoch 51/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1013 - val_loss: 0.1486\n",
            "Epoch 52/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1015 - val_loss: 0.1505\n",
            "Epoch 53/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1014 - val_loss: 0.1528\n",
            "Epoch 54/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1007 - val_loss: 0.1562\n",
            "Epoch 55/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1000 - val_loss: 0.1507\n",
            "Epoch 56/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.1000 - val_loss: 0.1510\n",
            "Epoch 57/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0998 - val_loss: 0.1568\n",
            "Epoch 58/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0996 - val_loss: 0.1478\n",
            "Epoch 59/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0988 - val_loss: 0.1452\n",
            "Epoch 60/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.1003 - val_loss: 0.1491\n",
            "Epoch 61/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0994 - val_loss: 0.1537\n",
            "Epoch 62/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0988 - val_loss: 0.1486\n",
            "Epoch 63/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0981 - val_loss: 0.1598\n",
            "Epoch 64/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0986 - val_loss: 0.1524\n",
            "Epoch 65/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0982 - val_loss: 0.1506\n",
            "Epoch 66/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0977 - val_loss: 0.1537\n",
            "Epoch 67/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0982 - val_loss: 0.1501\n",
            "Epoch 68/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0979 - val_loss: 0.1521\n",
            "Epoch 69/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0975 - val_loss: 0.1575\n",
            "Epoch 70/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0975 - val_loss: 0.1571\n",
            "Epoch 71/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0943 - val_loss: 0.1436\n",
            "Epoch 72/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0968 - val_loss: 0.1613\n",
            "Epoch 73/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0963 - val_loss: 0.1488\n",
            "Epoch 74/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0959 - val_loss: 0.1464\n",
            "Epoch 75/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0967 - val_loss: 0.1484\n",
            "Epoch 76/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0958 - val_loss: 0.1458\n",
            "Epoch 77/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0964 - val_loss: 0.1478\n",
            "Epoch 78/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0956 - val_loss: 0.1580\n",
            "Epoch 79/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0953 - val_loss: 0.1516\n",
            "Epoch 80/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0957 - val_loss: 0.1460\n",
            "Epoch 81/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0949 - val_loss: 0.1544\n",
            "Epoch 82/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0951 - val_loss: 0.1467\n",
            "Epoch 83/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0959 - val_loss: 0.1473\n",
            "Epoch 84/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0952 - val_loss: 0.1451\n",
            "Epoch 85/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0952 - val_loss: 0.1493\n",
            "Epoch 86/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0943 - val_loss: 0.1471\n",
            "Epoch 87/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0952 - val_loss: 0.1560\n",
            "Epoch 88/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0954 - val_loss: 0.1511\n",
            "Epoch 89/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0927 - val_loss: 0.1448\n",
            "Epoch 90/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0952 - val_loss: 0.1500\n",
            "Epoch 91/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0947 - val_loss: 0.1497\n",
            "Epoch 92/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0939 - val_loss: 0.1482\n",
            "Epoch 93/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0945 - val_loss: 0.1488\n",
            "Epoch 94/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0942 - val_loss: 0.1472\n",
            "Epoch 95/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0938 - val_loss: 0.1460\n",
            "Epoch 96/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0942 - val_loss: 0.1469\n",
            "Epoch 97/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0940 - val_loss: 0.1466\n",
            "Epoch 98/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0943 - val_loss: 0.1557\n",
            "Epoch 99/100\n",
            "40/40 [==============================] - 0s 1ms/step - loss: 0.0941 - val_loss: 0.1496\n",
            "Epoch 100/100\n",
            "40/40 [==============================] - 0s 2ms/step - loss: 0.0941 - val_loss: 0.1519\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NhUCLU45lTby",
        "outputId": "4b044f03-285a-4a90-eee9-924fcb007835",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "predictions = wider_model_rmsprop.predict(feature_test)\n",
        "predictions = predictions.flatten()\n",
        "\n",
        "mse = mean_squared_error(label_test, predictions)\n",
        "mae = mean_absolute_error(label_test, predictions)\n",
        "r2 = r2_score(label_test, predictions)\n",
        "print(\"MSE (Mean Squared Error)\", mse)\n",
        "print(\"MAE (Mean Absolute Error)\", mae)\n",
        "print(\"r^2 score\", r2)\n",
        "print('RMSE (Root Mean Squared Error', np.sqrt(mean_squared_error(label_test, predictions)))"
      ],
      "execution_count": 355,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:5 out of the last 13 calls to <function Model.make_predict_function.<locals>.predict_function at 0x7efbe39adc80> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/tutorials/customization/performance#python_or_tensor_args and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n",
            "MSE (Mean Squared Error) 0.15187030606009527\n",
            "MAE (Mean Absolute Error) 0.2987786783283517\n",
            "r^2 score 0.8486732511957779\n",
            "RMSE (Root Mean Squared Error 0.3897054093287586\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OwWCSGjLlfJx",
        "outputId": "36501cab-c13d-4eec-b2ad-761351814bf0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        }
      },
      "source": [
        "history_rmsprop_dataframe = pd.DataFrame(history_rmsprop.history)\n",
        "history_rmsprop_dataframe['epoch'] = history_rmsprop.epoch\n",
        "history_rmsprop_dataframe.sort_values(by='val_loss', ascending=True)"
      ],
      "execution_count": 356,
      "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>loss</th>\n",
              "      <th>val_loss</th>\n",
              "      <th>epoch</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>70</th>\n",
              "      <td>0.094279</td>\n",
              "      <td>0.143615</td>\n",
              "      <td>70</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>88</th>\n",
              "      <td>0.092661</td>\n",
              "      <td>0.144836</td>\n",
              "      <td>88</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>83</th>\n",
              "      <td>0.095247</td>\n",
              "      <td>0.145147</td>\n",
              "      <td>83</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>58</th>\n",
              "      <td>0.098786</td>\n",
              "      <td>0.145181</td>\n",
              "      <td>58</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>75</th>\n",
              "      <td>0.095760</td>\n",
              "      <td>0.145842</td>\n",
              "      <td>75</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0.198276</td>\n",
              "      <td>0.220871</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.222536</td>\n",
              "      <td>0.240643</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.291090</td>\n",
              "      <td>0.265540</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.525870</td>\n",
              "      <td>0.392765</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.842920</td>\n",
              "      <td>0.693474</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>100 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "        loss  val_loss  epoch\n",
              "70  0.094279  0.143615     70\n",
              "88  0.092661  0.144836     88\n",
              "83  0.095247  0.145147     83\n",
              "58  0.098786  0.145181     58\n",
              "75  0.095760  0.145842     75\n",
              "..       ...       ...    ...\n",
              "4   0.198276  0.220871      4\n",
              "3   0.222536  0.240643      3\n",
              "2   0.291090  0.265540      2\n",
              "1   0.525870  0.392765      1\n",
              "0   0.842920  0.693474      0\n",
              "\n",
              "[100 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 356
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3ShQQbQ1lfJ0",
        "outputId": "02bd5e33-25fa-4aca-d2c3-40d3c407eef4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        }
      },
      "source": [
        "plot_loss(history_rmsprop) # epoch vs loss graph"
      ],
      "execution_count": 357,
      "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": "markdown",
      "metadata": {
        "id": "xfnhC4qznqCV"
      },
      "source": [
        "# **Kesimpulan**"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tYZVgJ8Dntfz"
      },
      "source": [
        "\n",
        "```\n",
        "linear regression r^2 score 0.792478802889181\n",
        "Base Model DL r^2 score 0.8428530241835525\n",
        "Deeper Model DL r^2 score 0.842879213761671\n",
        "Wider Model DL r^2 score 0.8520116387147619\n",
        "opt SGD r^2 score 0.8496068113206386\n",
        "opt RMSProp r^2 score 0.8486732511957779\n",
        "```\n",
        "\n",
        "Menurut hasil sekali run diatas Hasil yang di dapat untuk r^2 terbaik ada pada wider model dengan optimizer Adam "
      ]
    }
  ]
}