(DSSG_WRI)_DHS_Analysis_.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "(DSSG/WRI) DHS Analysis .ipynb",
      "provenance": [],
      "toc_visible": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 137
        },
        "id": "DTGSvQn513aV",
        "outputId": "74810ba8-a2de-49e6-f76f-767ef568c8dd"
      },
      "source": [
        "# ignore\n",
        "# Was trying to read raw data stored in sav files\n",
        "'''\n",
        "!pip install pyreadstat\n",
        "import pandas as pd\n",
        "df = pd.read_spss('/content/drive/MyDrive/WRI\\DSSG - EDA/House Member Recode 2015.SAV')\n",
        "#df = pd.read_spss('C:\\\\Users\\\\rohaan.nadeem\\\\Desktop\\\\House Member Recode 2015.SAV')\n",
        "import pyreadstat\n",
        "fpath = '/content/drive/MyDrive/WRI\\DSSG - EDA/House Member Recode 2015.SAV'\n",
        "reader = pyreadstat.read_file_in_chunks(pyreadstat.read_sas7bdat, fpath, chunksize= 10000)\n",
        "\n",
        "for df, meta in reader:\n",
        "  print(df) # df will contain 10K rows\n",
        "\n",
        "!pip install savReaderWriter\n",
        "import savReaderWriter as spss\n",
        "import pandas as pd\n",
        "dataFile = '/content/drive/MyDrive/WRI\\DSSG - EDA/House Member Recode 2015.SAV'\n",
        "raw_data = spss.SavReader(dataFile, returnHeader = True) # This is fast\n",
        "data = pd.DataFrame(raw_data)\n",
        "'''"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "\"\\n!pip install pyreadstat\\nimport pandas as pd\\ndf = pd.read_spss('/content/drive/MyDrive/WRI\\\\DSSG - EDA/House Member Recode 2015.SAV')\\n#df = pd.read_spss('C:\\\\Users\\\\rohaan.nadeem\\\\Desktop\\\\House Member Recode 2015.SAV')\\nimport pyreadstat\\nfpath = '/content/drive/MyDrive/WRI\\\\DSSG - EDA/House Member Recode 2015.SAV'\\nreader = pyreadstat.read_file_in_chunks(pyreadstat.read_sas7bdat, fpath, chunksize= 10000)\\n\\nfor df, meta in reader:\\n  print(df) # df will contain 10K rows\\n\\n!pip install savReaderWriter\\nimport savReaderWriter as spss\\nimport pandas as pd\\ndataFile = '/content/drive/MyDrive/WRI\\\\DSSG - EDA/House Member Recode 2015.SAV'\\nraw_data = spss.SavReader(dataFile, returnHeader = True) # This is fast\\ndata = pd.DataFrame(raw_data)\\n\""
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 2
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kBOU4GnV33Yp"
      },
      "source": [
        "External Libraries"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Sni866Rc33Gb"
      },
      "source": [
        "import numpy as np # for linear algebra\n",
        "import pandas as pd # for dataframes\n",
        "import matplotlib.pyplot as plt # for data visiualization\n",
        "import seaborn as sns # for data visualization\n",
        "from sklearn import preprocessing # for one hot encoding\n",
        "from sklearn.pipeline import Pipeline # for pipelines\n",
        "from sklearn.model_selection import train_test_split # for tts\n",
        "from sklearn import svm # classification model\n",
        "from sklearn.metrics import mean_absolute_error, jaccard_score, f1_score, log_loss, accuracy_score # metrics\n",
        "from sklearn import metrics # for accuracy score metric\n",
        "from sklearn.tree import DecisionTreeClassifier # Decision Trees\n",
        "from tensorflow import keras # for deep learning\n",
        "from tensorflow.keras import layers # for deep learning\n",
        "%matplotlib inline "
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YL5U9R2a4lxw"
      },
      "source": [
        "Read data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BPZBPlJGZ5ei",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "fdb077c6-3052-48dc-a18c-e752256791a8"
      },
      "source": [
        "df = pd.read_csv('/content/drive/MyDrive/WRI\\DSSG - EDA/DHS-PROCESSED-CLEAN.csv',\n",
        "                 #index_col = 'DHSCLUST'\t\n",
        "                 )\n",
        "df.shape"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(28385, 16)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "k4vGCmPaabVS",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 411
        },
        "outputId": "35c7c230-abe2-4f87-8cb1-128718522975"
      },
      "source": [
        "# glimpse data\n",
        "df.head()"
      ],
      "execution_count": 5,
      "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>DHSCLUST</th>\n",
              "      <th>Wealth</th>\n",
              "      <th>Wealth Index</th>\n",
              "      <th>Toilet Facility</th>\n",
              "      <th>Roof Material</th>\n",
              "      <th>Electricity</th>\n",
              "      <th>Cooking Fuel</th>\n",
              "      <th>Drinking Water</th>\n",
              "      <th>URBAN_RURA</th>\n",
              "      <th>DHSREGNA</th>\n",
              "      <th>ADM1NAME</th>\n",
              "      <th>LATNUM</th>\n",
              "      <th>LONGNUM</th>\n",
              "      <th>geometry</th>\n",
              "      <th>UN_Population_Density_2015</th>\n",
              "      <th>Population_Density</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>10001</td>\n",
              "      <td>richer</td>\n",
              "      <td>107648.636364</td>\n",
              "      <td>Flush</td>\n",
              "      <td>Metal</td>\n",
              "      <td>yes</td>\n",
              "      <td>Petroleum</td>\n",
              "      <td>Piped Water</td>\n",
              "      <td>U</td>\n",
              "      <td>South Andaman</td>\n",
              "      <td>Andaman &amp; Nicobar Islands</td>\n",
              "      <td>11.670645</td>\n",
              "      <td>92.745799</td>\n",
              "      <td>POINT (92.74579928369999 11.6706448114)</td>\n",
              "      <td>371.654433</td>\n",
              "      <td>Average</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>10002</td>\n",
              "      <td>middle</td>\n",
              "      <td>25279.409091</td>\n",
              "      <td>Flush</td>\n",
              "      <td>Metal</td>\n",
              "      <td>yes</td>\n",
              "      <td>Petroleum</td>\n",
              "      <td>Piped Water</td>\n",
              "      <td>R</td>\n",
              "      <td>Nicobars</td>\n",
              "      <td>Andaman &amp; Nicobar Islands</td>\n",
              "      <td>9.143844</td>\n",
              "      <td>92.826752</td>\n",
              "      <td>POINT (92.8267516352 9.143843815149999)</td>\n",
              "      <td>129.322150</td>\n",
              "      <td>Very Low</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>10003</td>\n",
              "      <td>middle</td>\n",
              "      <td>4725.681818</td>\n",
              "      <td>Flush</td>\n",
              "      <td>Metal</td>\n",
              "      <td>yes</td>\n",
              "      <td>Wood/Straw</td>\n",
              "      <td>Piped Water</td>\n",
              "      <td>R</td>\n",
              "      <td>Nicobars</td>\n",
              "      <td>Andaman &amp; Nicobar Islands</td>\n",
              "      <td>9.212246</td>\n",
              "      <td>92.753633</td>\n",
              "      <td>POINT (92.75363284469999 9.212245586050001)</td>\n",
              "      <td>129.322140</td>\n",
              "      <td>Very Low</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>10004</td>\n",
              "      <td>middle</td>\n",
              "      <td>17484.909091</td>\n",
              "      <td>Flush</td>\n",
              "      <td>Metal</td>\n",
              "      <td>yes</td>\n",
              "      <td>Wood/Straw</td>\n",
              "      <td>Piped Water</td>\n",
              "      <td>R</td>\n",
              "      <td>Nicobars</td>\n",
              "      <td>Andaman &amp; Nicobar Islands</td>\n",
              "      <td>9.165413</td>\n",
              "      <td>92.742696</td>\n",
              "      <td>POINT (92.74269589719999 9.165413484529999)</td>\n",
              "      <td>129.322139</td>\n",
              "      <td>Very Low</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>10005</td>\n",
              "      <td>middle</td>\n",
              "      <td>25083.454545</td>\n",
              "      <td>Flush</td>\n",
              "      <td>Metal</td>\n",
              "      <td>yes</td>\n",
              "      <td>Petroleum</td>\n",
              "      <td>Piped Water</td>\n",
              "      <td>R</td>\n",
              "      <td>Nicobars</td>\n",
              "      <td>Andaman &amp; Nicobar Islands</td>\n",
              "      <td>8.307356</td>\n",
              "      <td>93.093792</td>\n",
              "      <td>POINT (93.0937924407 8.30735625773)</td>\n",
              "      <td>20.672477</td>\n",
              "      <td>Very Low</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   DHSCLUST  Wealth  ...  UN_Population_Density_2015 Population_Density\n",
              "0     10001  richer  ...                  371.654433            Average\n",
              "1     10002  middle  ...                  129.322150           Very Low\n",
              "2     10003  middle  ...                  129.322140           Very Low\n",
              "3     10004  middle  ...                  129.322139           Very Low\n",
              "4     10005  middle  ...                   20.672477           Very Low\n",
              "\n",
              "[5 rows x 16 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZCRRoMqG5IaW"
      },
      "source": [
        "Since wealth is the target, need to ensure the data is balanced."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "w-KNMxVEH-Os"
      },
      "source": [
        "## Data Balancing"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 340
        },
        "id": "bsTdTYOA8KOv",
        "outputId": "1c251626-03a4-4976-83e2-4888f7560798"
      },
      "source": [
        "df.Wealth.value_counts().plot.bar(title = 'Distribution of Wealth Classes',\n",
        "                                  xlabel = 'Wealth Class',\n",
        "                                  ylabel = 'Number of Records')"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fe6d4355c90>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 374
        },
        "id": "EB33vNBxV6qW",
        "outputId": "bfa1a46e-0af8-4c3d-d358-da3dc53eec5c"
      },
      "source": [
        "#Using matplotlib\n",
        "pie, ax = plt.subplots(figsize=[10,6])\n",
        "labels = df.Wealth.unique()\n",
        "plt.pie(x=df.Wealth.value_counts(), autopct=\"%.1f%%\", labels=labels, explode=[0.05]*5, pctdistance=0.5)\n",
        "plt.title(\"Wealth Label Distribution\", fontsize=14);\n",
        "#pie.savefig(\"DeliveryPieChart.png\")"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x432 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 340
        },
        "id": "EUIIXU398lok",
        "outputId": "5554408d-5290-41f1-e598-381b027ab917"
      },
      "source": [
        "wealth_records = df.Wealth.value_counts()\n",
        "wealth_records_unbalance = wealth_records - df.Wealth.value_counts().mean()\n",
        "wealth_records_unbalance.plot.bar( title = 'Class Records Imbalance',\n",
        "                                  xlabel = 'Wealth Class',\n",
        "                                  ylabel = 'Variance (Number of Records)')"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fe6d3b53110>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ujeEoRIC_Osp",
        "outputId": "fd538bd9-bc06-49d3-d79a-6ab94f378cda"
      },
      "source": [
        "print('Number of Poorest class records: ', wealth_records[0])\n",
        "print('Number of Richer class records: ', wealth_records[-1])\n",
        "print('Poorest to Richer Ratio: ', round(wealth_records[0]/wealth_records[-1],3))"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Number of Poorest class records:  7472\n",
            "Number of Richer class records:  4724\n",
            "Poorest to Richer Ratio:  1.582\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "c8aQ8die_7aM"
      },
      "source": [
        "Looks like data is highly unbalanced towards the poorest class.\n",
        "\n",
        "Removing bias via random sampling"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "McPHvyZn_MsD"
      },
      "source": [
        "'''# undersampling poorest class\n",
        "delta = wealth_records[0] - wealth_records[-1]\n",
        "to_remove = np.random.choice(df[df['Wealth'] == 'poorest'].index, delta ,replace=False)\n",
        "df.drop(to_remove, inplace = True)\n",
        "\n",
        "# undersampling poorer class\n",
        "delta = wealth_records[1] - wealth_records[-1]\n",
        "to_remove = np.random.choice(df[df['Wealth'] == 'poorer'].index, delta ,replace=False)\n",
        "df.drop(to_remove, inplace = True)\n",
        "\n",
        "# undersampling richest class\n",
        "delta = wealth_records[2] - wealth_records[-1]\n",
        "to_remove = np.random.choice(df[df['Wealth'] == 'richest'].index, delta ,replace=False)\n",
        "df.drop(to_remove, inplace = True)\n",
        "\n",
        "# undersampling poorer class\n",
        "delta = wealth_records[3] - wealth_records[-1]\n",
        "to_remove = np.random.choice(df[df['Wealth'] == 'middle'].index, delta ,replace=False)\n",
        "df.drop(to_remove, inplace = True)\n",
        "\n",
        "df.Wealth.value_counts().plot.bar(title = 'Distribution of Wealth Classes (random undersampled)',\n",
        "                                  xlabel = 'Wealth Class',\n",
        "                                  ylabel = 'Number of Records')\n",
        "\n",
        "'''"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nhTJza6WH1__"
      },
      "source": [
        "## EDA"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fugfxt7SAF7u"
      },
      "source": [
        "For proper visualization, understanding data types"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pc5jTaUx4mGU"
      },
      "source": [
        "df.dtypes"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vtMudzWp5MiP"
      },
      "source": [
        "#### Quantitative Analysis"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GCSsvdbhHyJW"
      },
      "source": [
        "df.describe()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZweasTee2xC3"
      },
      "source": [
        "Questions:\n",
        "1. Does Wealth Index has outliers? (Box and whiskers will tell)."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MntzY5ps3CBm"
      },
      "source": [
        "df.boxplot(column = ['Wealth Index'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "loiyWa_04DVE"
      },
      "source": [
        "Looks okay. \n",
        "\n",
        "What about correlations?"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fL_2Z0nzAl_k"
      },
      "source": [
        "df.corr()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xIy37sL0BDOD"
      },
      "source": [
        "*   Latitude and Longitude are irrelevant. \n",
        "*   Population density has a weak correlation with Wealth Index. Makes sense. \n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fPyaIfi-5nt5"
      },
      "source": [
        "#### Descriptive Analysis"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g7HnU03t5aGS"
      },
      "source": [
        "df.describe(include = 'object')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ejJ9cJRU532h"
      },
      "source": [
        "Observations:\n",
        "*   No missing data\n",
        "\n",
        "Questions:\n",
        "\n",
        "*   Relation between Toilet Facilitty & Wealth\n",
        "*   Relation between Roof Material & Wealth\n",
        "*   Relation between Electricity & Wealth\n",
        "*   Relation between Drinking Water & Wealth\n",
        "*   Relation between Drinking Population Density & Wealth\n",
        "\n",
        "The columns I think will not help:\n",
        "\n",
        "*   ~Urban/Rural~\n",
        "*   DHSREGNA\t\n",
        "*   ADM1NAME\t\n",
        "*   geometry\n",
        "*   LATNUM\n",
        "*   LONGNUM\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sXfbh2hb3IBJ"
      },
      "source": [
        "df.drop(columns = ['DHSREGNA', 'ADM1NAME', 'LATNUM', 'LONGNUM', 'geometry'],\n",
        "        inplace = True)\n",
        "df.shape"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4s8vQ1PH_ABv"
      },
      "source": [
        "df.columns"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1TSi0FsuOoro"
      },
      "source": [
        "##### Without Rural/Urban Flag"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NlHDtNICCTfh"
      },
      "source": [
        "Relation between Toilet Facilitty & Wealth"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9meQ4TTKCMPw"
      },
      "source": [
        "sns.set(rc={'figure.figsize':(20, 10)})\n",
        "sns.boxplot(x=\"Toilet Facility\", \n",
        "            y=\"Wealth Index\", \n",
        "            data=df)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xTZGDOakCtuO"
      },
      "source": [
        "Relation between Roof Material & Wealth"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "D16OmQDECtOH"
      },
      "source": [
        "sns.set(rc={'figure.figsize':(20, 10)})\n",
        "\n",
        "sns.boxplot(x=\"Roof Material\", \n",
        "            y=\"Wealth Index\",  \n",
        "            data=df)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MmpaVtgGECTr"
      },
      "source": [
        "Relation between Electricity & Wealth"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BYIl9_YOCkN5"
      },
      "source": [
        "sns.set(rc={'figure.figsize':(20, 10)})\n",
        "\n",
        "sns.boxplot(x=\"Electricity\", \n",
        "            y=\"Wealth Index\",  \n",
        "            data=df)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Oj5sbYnkEsxR"
      },
      "source": [
        "Relation between Drinking Water & Wealth"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mws9JTDXC2yd"
      },
      "source": [
        "sns.set(rc={'figure.figsize':(20, 10)})\n",
        "\n",
        "sns.boxplot(x=\"Drinking Water\", \n",
        "            y=\"Wealth Index\",  \n",
        "            data=df)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gxIWihpGFKYZ"
      },
      "source": [
        "I think differentiating between Rurual and Urban will help."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fFDiZDdZFXco"
      },
      "source": [
        "Relation between Drinking Population Density & Wealth"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6NWFpYboE0CT"
      },
      "source": [
        "sns.set(rc={'figure.figsize':(20, 10)})\n",
        "\n",
        "sns.boxplot(x=\"Population_Density\", \n",
        "            y=\"Wealth Index\",  \n",
        "            data=df)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EGwDXslaOtRH"
      },
      "source": [
        "##### Grouped by Rural/Urban"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4V1sAzyrV_av"
      },
      "source": [
        "Relation between Toilet Facilitty & Wealth"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "W57IdDPoV_bC"
      },
      "source": [
        "sns.boxplot(x=\"Toilet Facility\", \n",
        "            y=\"Wealth Index\",\n",
        "            hue='URBAN_RURA', \n",
        "            data=df)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "x3m908c1V_bE"
      },
      "source": [
        "Relation between Roof Material & Wealth"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "diDFsRUSV_bF"
      },
      "source": [
        "sns.set(rc={'figure.figsize':(20, 10)})\n",
        "\n",
        "sns.boxplot(x=\"Roof Material\", \n",
        "            y=\"Wealth Index\",\n",
        "            hue='URBAN_RURA',  \n",
        "            data=df)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XJ8kGapTV_bF"
      },
      "source": [
        "Relation between Electricity & Wealth"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9kx44PhLV_bG"
      },
      "source": [
        "sns.set(rc={'figure.figsize':(20, 10)})\n",
        "\n",
        "sns.boxplot(x=\"Electricity\", \n",
        "            y=\"Wealth Index\",\n",
        "            hue='URBAN_RURA',  \n",
        "            data=df)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0ZvxIF27V_bH"
      },
      "source": [
        "Relation between Drinking Water & Wealth"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "L6NnJrGRV_bH"
      },
      "source": [
        "sns.set(rc={'figure.figsize':(20, 10)})\n",
        "\n",
        "sns.boxplot(x=\"Drinking Water\", \n",
        "            y=\"Wealth Index\",\n",
        "            hue='URBAN_RURA',  \n",
        "            data=df)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1KqVff47V_bI"
      },
      "source": [
        "Relation between Drinking Population Density & Wealth"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h7UQlMjtV_bJ"
      },
      "source": [
        "sns.set(rc={'figure.figsize':(5, 4)})\n",
        "\n",
        "sns.boxplot(x=\"Population_Density\", \n",
        "            y=\"Wealth Index\",  \n",
        "            data=df)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ipKVvMnjbYrq"
      },
      "source": [
        "## Modelling & Evaluation"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wxrFU6e-bIna"
      },
      "source": [
        "\n",
        "This is a classification problem. Classification algorithms that might work:\n",
        "\n",
        "\n",
        "\n",
        "*   SVM\n",
        "*   Logistic Regression\n",
        "*   Decision Trees\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5AcNSLmCbtR4"
      },
      "source": [
        "### Preprocessing"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6uaL1nK4cFLz"
      },
      "source": [
        "df.columns"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "z_FNihMTdsJo"
      },
      "source": [
        "X = df[['Toilet Facility', \n",
        "       'Roof Material', \n",
        "       'Electricity', \n",
        "       'Cooking Fuel', \n",
        "       'Drinking Water', \n",
        "       'URBAN_RURA', \n",
        "       'Population_Density']].values\n",
        "    \n",
        "y = df['Wealth']\n",
        "\n",
        "y[0:5]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RYTzj4_al158"
      },
      "source": [
        "le_toilet_facility = preprocessing.LabelEncoder()\n",
        "le_toilet_facility.fit(['Flush', 'No Facility', 'Pit Latrine', 'Other'])\n",
        "X[:,0] = le_toilet_facility.transform(X[:,0]) \n",
        "\n",
        "le_roof_material = preprocessing.LabelEncoder()\n",
        "le_roof_material.fit(['Metal', 'Cement', 'Stones', 'Thatch/Palm', 'Tiles', 'Others', 'Mud', 'Bricks', 'Mat/Sheets', 'Wood', 'No Roof'])\n",
        "X[:,1] = le_roof_material.transform(X[:,1])\n",
        "\n",
        "le_electricity = preprocessing.LabelEncoder()\n",
        "le_electricity.fit(['yes', 'no'])\n",
        "X[:,2] = le_electricity.transform(X[:,2]) \n",
        "\n",
        "le_cooking_fuel = preprocessing.LabelEncoder()\n",
        "le_cooking_fuel.fit(['Petroleum', 'Wood/Straw', 'Electricity', 'Crops', 'Dung', 'Biogas', 'Coal'])\n",
        "X[:,3] = le_cooking_fuel.transform(X[:,3])\n",
        "\n",
        "le_drinking_water = preprocessing.LabelEncoder()\n",
        "le_drinking_water.fit(['Piped Water', 'Public Tap', 'Tube Well/Borehole', 'Well', 'Bottled', 'River/Dam/Lake', 'Ro Plant', 'Tanker', 'Spring', 'Other', 'Small Tank', 'Rainwater'])\n",
        "X[:,4] = le_drinking_water.transform(X[:,4])\n",
        "\n",
        "le_urban_rural = preprocessing.LabelEncoder()\n",
        "le_urban_rural.fit(['U', 'R'])\n",
        "X[:,5] = le_urban_rural.transform(X[:,5])\n",
        "\n",
        "le_population_density = preprocessing.LabelEncoder()\n",
        "le_population_density.fit(['Average', 'Very Low', 'Low', 'High', 'Very High'])\n",
        "X[:,6] = le_population_density.transform(X[:,6])\n",
        "\n",
        "X[:5]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UX2fEIRkr_xl"
      },
      "source": [
        "y.unique()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-9BX1jRHr367"
      },
      "source": [
        "le_wealth = preprocessing.LabelEncoder()\n",
        "le_wealth.fit(['richer', 'middle', 'poorest', 'poorer', 'richest'])\n",
        "y = le_wealth.transform(y)\n",
        "y[:5]\n",
        "#le_wealth"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XwHF61YMbvb1"
      },
      "source": [
        "# test-train split\n",
        "X_trainset, X_testset, y_trainset, y_testset = train_test_split(X, y, test_size=0.3, random_state=3)\n",
        "print('Shape of X_trainset :',X_trainset.shape,'\\nShape of y_trainset :',y_trainset.shape)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ABH3AH8bqcMn"
      },
      "source": [
        "X_trainset"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HQS2s8KDg9sH"
      },
      "source": [
        "# to measure performance\n",
        "def evaluate_ai (y_testset, y_hat):\n",
        "    print(\"AI's Accuracy: \", round(metrics.accuracy_score(y_testset, y_hat),2))\n",
        "    print(\"AI's Jaccard Similarity Index: \", round(jaccard_score(y_testset, y_hat, average='weighted'),2))\n",
        "    print(\"AI's F1 Score : \", round(f1_score(y_testset, y_hat, average='weighted'),2))\n",
        "    #print(\"AI's Log Loss: \", round(log_loss(y_testset, y_hat),2))\n",
        "    print(\"AI's MAE: \", round(mean_absolute_error(y_testset, y_hat),2))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CuV8J400bqIR"
      },
      "source": [
        "### SVM"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qxKu5nb4gU4H"
      },
      "source": [
        "clf = svm.SVC(kernel='linear', gamma = 'auto')\n",
        "clf.fit(X_trainset,y_trainset)\n",
        "y_hat = clf.predict(X_testset)  \n",
        "evaluate_ai(y_testset, y_hat)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SlSOlVPxuHkn"
      },
      "source": [
        "mean_absolute_error(y_hat, y_testset)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "X5wEm2f9Kasn"
      },
      "source": [
        "### Decision Tree"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l3FNHt_c_U9P"
      },
      "source": [
        "\n",
        "Ks = 12\n",
        "mean_acc= np.zeros((Ks-1))\n",
        "std_acc = np.zeros((Ks-1))\n",
        "ConfustionMx = [];\n",
        "for n in range(1,Ks):\n",
        "    \n",
        "    #Train Model and Predict  \n",
        "    k_hunter = DecisionTreeClassifier(criterion=\"entropy\", max_depth = n).fit(X_trainset,y_trainset)\n",
        "    y_hat2=k_hunter.predict(X_testset)\n",
        "    mean_acc[n-1] = accuracy_score(y_testset, y_hat2)\n",
        "    std_acc[n-1]=np.std(y_hat2==y_testset)/np.sqrt(y_hat2.shape[0])\n",
        "\n",
        "plt.plot(range(1,Ks),mean_acc,'g')\n",
        "plt.fill_between(range(1,Ks),mean_acc - 1 * std_acc,mean_acc + 1 * std_acc, alpha=0.10)\n",
        "plt.legend(('Accuracy Score ', '+/- 3xstd'))\n",
        "plt.ylabel('Accuracy Score')\n",
        "plt.xlabel('Depth')\n",
        "plt.tight_layout()\n",
        "plt.show()\n",
        "\n",
        "Tree_optimizers = []\n",
        "for i in range(0,len(mean_acc)):\n",
        "    if mean_acc[i] == mean_acc.max():\n",
        "        Tree_optimizers.append(i)\n",
        "        print('Depth With Max Accuracy :',i,'\\nMax accuracy :',mean_acc.max())\n",
        "\n",
        "print('\\n')\n",
        "depth_optimized = Tree_optimizers[0]\n",
        "Tree_optimized = DecisionTreeClassifier(criterion=\"entropy\", max_depth = depth_optimized)\n",
        "Tree_optimized.fit(X_trainset,y_trainset)\n",
        "y_hat_optimized = Tree_optimized.predict(X_testset)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Bs7-tyebA4gp"
      },
      "source": [
        "# Build the model\n",
        "DTree = DecisionTreeClassifier(criterion=\"entropy\", max_depth = 9)\n",
        "DTree # it shows the default parameters\n",
        "\n",
        "# Fit the model\n",
        "DTree.fit(X_trainset,y_trainset)\n",
        "\n",
        "# Make predictions\n",
        "predTree = DTree.predict(X_testset)\n",
        "evaluate_ai (y_testset, predTree)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-POxwd8fKen5"
      },
      "source": [
        "### Classification via Neural Network"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3xMqD-RVKh2p"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NxWuc6BWrtB0"
      },
      "source": [
        "## Best Performing Models:\n",
        "\n",
        "| Approach | Mean Absolute Error |\n",
        "|:--------:|:-------------------:|\n",
        "| Linear Regression With One Hot Encoding | 27638.585 |\n",
        "| Linear Lasso Regression With Catboost Encoder | 28267.662 |\n",
        "| Decision Tree Regression (depth = 5) <br> With CatBoost Encoder | 28096.050 |\n",
        "| Decision Tree Regression (depth = 8) <br> With One Hot Encoding | 26523.034 | \n",
        "| CatBoost Regression With CatBoost Encoder | 26392.809 |\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "i_IegMSIrtnk"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}