NN Classification Project.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "7por6hHzcgTW"
   },
   "source": [
    "# Individual NN Classification Assignment (Spring 2024)\n",
    "**Dr. Dave Wanik - University of Connecticut**\n",
    "\n",
    "* Full Name: Adam Syed\n",
    "* Student ID (7 digit number): 2735566\n",
    "\n",
    "---------------------------------------------------------------------------\n",
    "\n",
    "Each student must complete and submit their own file, but you may work with your group for help. I do not want to see direct copy and pasted code otherwise you will get a zero.\n",
    "\n",
    "1. Read the data dictionary and description from here:\n",
    "\n",
    "* https://oehha.ca.gov/media/downloads/calenviroscreen/document/calenviroscreen40resultsdatadictionaryf2021.zip\n",
    "\n",
    " Download the data using gdown, read the Excel file using pandas, print the first 10 rows using df.head() and use df.info() to examine the data types and missing values.\n",
    "\n",
    "2. Simplify the raw dataframe so that you only keep the columns you need. The `X` variable will be the following columns: `Population`, `Ozone` through `Solid Waste Pctl`, and `Asthma` through `Linguistic Isolation Pctl`. The `y` variable will be `Poverty`. Examine the quality of each column and use your judgement about dropping rows or imputing missing values. Add text cells and lots of comments so we can understand your logic/justification!\n",
    "\n",
    "3. Recode the target variable to a 1 if greater than the mean value of poverty, otherwise make it a 0. Use this recoded variable as the target variable! Now it is a classification problem.\n",
    "\n",
    "4. Make two interesting plots or tables and a description of why you made the table and what you see.\n",
    "\n",
    "5. Do an 90/10 split for X_train, X_test, y_train, y_test where the random seed is equal to your 7 digit studentID number.\n",
    "\n",
    "6. Use the StandardScaler() on train and apply to test partition. Do not scale the target variable!\n",
    "\n",
    "7. Build a model using the Sequential API (like we do in class) with at least 2 dense layers with the relu activation function, and with dropout in between each dense layer (use a number between 0.1 and 0.5). Compile the model using an appropriate optimizer. Use early stopping with patience of at least 10 and restore the best weights once the model converges. You can choose whatever batch size you would like to.\n",
    "\n",
    "8. Fit the model for 100000 epochs with a batch size of your choice, using X_test and y_test as the validation data. **Don’t forget the early stopping callback!**\n",
    "\n",
    "9. Evaluate the model using learning curves, error metrics and confusion matrices for each partition (like we do in class). You should largely be able to copy and paste this from class notebooks. Add a few bullet points about what you see (did your model learn nice and gently?  If you don't have text cells here, you will lose points.\n",
    "\n",
    "10. Calculate what a baseline prediction would be for the train and test partitions (a mean only model). Did your model do better than the baseline predictions? If so, you have a useful model!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "RSBDw0iEErUJ"
   },
   "source": [
    "# Read Data and Import Modules\n",
    "## Question 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "metadata": {
    "id": "gFcKcj142-Yw"
   },
   "outputs": [],
   "source": [
    "# import modules\n",
    "# for general data analysis/plotting\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "\n",
    "# for data prep\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "# neural net modules\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout\n",
    "from keras.callbacks import EarlyStopping"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "XKQYozqn3Drz",
    "outputId": "2ca8e535-ddd7-4e44-9ce7-c60479ea0f19"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/Users/adamyeeter/anaconda3/lib/python3.11/site-packages/gdown/cli.py:121: FutureWarning: Option `--id` was deprecated in version 4.3.1 and will be removed in 5.0. You don't need to pass it anymore to use a file ID.\n",
      "  warnings.warn(\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1_8vmQwSZ02ZOMw_IPHY5OWkwpXQLVW2E\n",
      "To: /Users/adamyeeter/OPIM 5509 Deep Learning/M3 Assignment/CalEnviroScreen.xlsx\n",
      "100%|██████████████████████████████████████| 6.19M/6.19M [00:00<00:00, 15.5MB/s]\n"
     ]
    }
   ],
   "source": [
    "# https://drive.google.com/file/d/1_8vmQwSZ02ZOMw_IPHY5OWkwpXQLVW2E/view?usp=sharing\n",
    "!gdown --id 1_8vmQwSZ02ZOMw_IPHY5OWkwpXQLVW2E # ID for Environmental Health data! look up!\n",
    "df_CalEnviroScreen = pd.read_excel('CalEnviroScreen.xlsx')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "6STrxO8F3J1r",
    "outputId": "e915dbf9-9a2e-4f0a-e2f5-f219d43f89dd"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 8035 entries, 0 to 8034\n",
      "Data columns (total 57 columns):\n",
      " #   Column                                            Non-Null Count  Dtype  \n",
      "---  ------                                            --------------  -----  \n",
      " 0   Census Tract                                      8035 non-null   int64  \n",
      " 1   Total Population                                  8035 non-null   int64  \n",
      " 2   California County                                 8035 non-null   object \n",
      " 3   ZIP                                               8035 non-null   int64  \n",
      " 4   Nearby City \n",
      "(to help approximate location only)  8035 non-null   object \n",
      " 5   Longitude                                         8035 non-null   float64\n",
      " 6   Latitude                                          8035 non-null   float64\n",
      " 7   CES 3.0 Score                                     7929 non-null   float64\n",
      " 8    CES 3.0 Percentile                               7929 non-null   float64\n",
      " 9   CES 3.0 \n",
      "Percentile Range                         7929 non-null   object \n",
      " 10  SB 535 Disadvantaged Community                    8035 non-null   object \n",
      " 11  Ozone                                             8035 non-null   float64\n",
      " 12  Ozone Pctl                                        8035 non-null   float64\n",
      " 13  PM2.5                                             8016 non-null   float64\n",
      " 14  PM2.5 Pctl                                        8016 non-null   float64\n",
      " 15  Diesel PM                                         8035 non-null   float64\n",
      " 16  Diesel PM Pctl                                    8035 non-null   float64\n",
      " 17  Drinking Water                                    8017 non-null   float64\n",
      " 18  Drinking Water Pctl                               8017 non-null   float64\n",
      " 19  Pesticides                                        8035 non-null   float64\n",
      " 20  Pesticides Pctl                                   8035 non-null   float64\n",
      " 21  Tox. Release                                      8035 non-null   float64\n",
      " 22  Tox. Release Pctl                                 8035 non-null   float64\n",
      " 23  Traffic                                           7979 non-null   float64\n",
      " 24  Traffic Pctl                                      7979 non-null   float64\n",
      " 25  Cleanup Sites                                     8035 non-null   float64\n",
      " 26  Cleanup Sites Pctl                                8035 non-null   float64\n",
      " 27  Groundwater Threats                               8035 non-null   float64\n",
      " 28  Groundwater Threats Pctl                          8035 non-null   float64\n",
      " 29  Haz. Waste                                        8035 non-null   float64\n",
      " 30  Haz. Waste Pctl                                   8035 non-null   float64\n",
      " 31  Imp. Water Bodies                                 8035 non-null   int64  \n",
      " 32  Imp. Water Bodies Pctl                            8035 non-null   float64\n",
      " 33  Solid Waste                                       8035 non-null   float64\n",
      " 34  Solid Waste Pctl                                  8035 non-null   float64\n",
      " 35  Pollution Burden                                  8035 non-null   float64\n",
      " 36  Pollution Burden Score                            8035 non-null   float64\n",
      " 37  Pollution Burden Pctl                             8035 non-null   float64\n",
      " 38  Asthma                                            8035 non-null   float64\n",
      " 39  Asthma Pctl                                       8035 non-null   float64\n",
      " 40  Low Birth Weight                                  7813 non-null   float64\n",
      " 41  Low Birth Weight Pctl                             7813 non-null   float64\n",
      " 42  Cardiovascular Disease                            8035 non-null   float64\n",
      " 43  Cardiovascular Disease Pctl                       8035 non-null   float64\n",
      " 44  Education                                         7939 non-null   float64\n",
      " 45  Education Pctl                                    7939 non-null   float64\n",
      " 46  Linguistic Isolation                              7793 non-null   float64\n",
      " 47  Linguistic Isolation Pctl                         7793 non-null   float64\n",
      " 48  Poverty                                           7956 non-null   float64\n",
      " 49  Poverty Pctl                                      7956 non-null   float64\n",
      " 50  Unemployment                                      7880 non-null   float64\n",
      " 51  Unemployment Pctl                                 7880 non-null   float64\n",
      " 52  Housing Burden                                    7878 non-null   float64\n",
      " 53  Housing Burden Pctl                               7878 non-null   float64\n",
      " 54  Pop. Char.                                        7929 non-null   float64\n",
      " 55  Pop. Char. Score                                  7929 non-null   float64\n",
      " 56  Pop. Char. Pctl                                   7929 non-null   float64\n",
      "dtypes: float64(49), int64(4), object(4)\n",
      "memory usage: 3.5+ MB\n"
     ]
    }
   ],
   "source": [
    "# read data\n",
    "df = pd.read_excel('CalEnviroScreen.xlsx')\n",
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 655
    },
    "id": "eCt8O7W85Y_z",
    "outputId": "f8362807-48a2-4fad-fd38-3191b5447f9f"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Census Tract</th>\n",
       "      <th>Total Population</th>\n",
       "      <th>California County</th>\n",
       "      <th>ZIP</th>\n",
       "      <th>Nearby City \\n(to help approximate location only)</th>\n",
       "      <th>Longitude</th>\n",
       "      <th>Latitude</th>\n",
       "      <th>CES 3.0 Score</th>\n",
       "      <th>CES 3.0 Percentile</th>\n",
       "      <th>CES 3.0 \\nPercentile Range</th>\n",
       "      <th>...</th>\n",
       "      <th>Linguistic Isolation Pctl</th>\n",
       "      <th>Poverty</th>\n",
       "      <th>Poverty Pctl</th>\n",
       "      <th>Unemployment</th>\n",
       "      <th>Unemployment Pctl</th>\n",
       "      <th>Housing Burden</th>\n",
       "      <th>Housing Burden Pctl</th>\n",
       "      <th>Pop. Char.</th>\n",
       "      <th>Pop. Char. Score</th>\n",
       "      <th>Pop. Char. Pctl</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6019001100</td>\n",
       "      <td>3174</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>93706</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>-119.781696</td>\n",
       "      <td>36.709695</td>\n",
       "      <td>94.090246</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>77.509665</td>\n",
       "      <td>76.3</td>\n",
       "      <td>97.121307</td>\n",
       "      <td>17.6</td>\n",
       "      <td>91.724838</td>\n",
       "      <td>26.0</td>\n",
       "      <td>79.398324</td>\n",
       "      <td>92.120494</td>\n",
       "      <td>9.553509</td>\n",
       "      <td>99.697314</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6071001600</td>\n",
       "      <td>6133</td>\n",
       "      <td>San Bernardino</td>\n",
       "      <td>91761</td>\n",
       "      <td>Ontario</td>\n",
       "      <td>-117.618013</td>\n",
       "      <td>34.057780</td>\n",
       "      <td>90.677839</td>\n",
       "      <td>99.987388</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>96.253833</td>\n",
       "      <td>72.5</td>\n",
       "      <td>94.632307</td>\n",
       "      <td>12.3</td>\n",
       "      <td>71.823836</td>\n",
       "      <td>34.1</td>\n",
       "      <td>93.754760</td>\n",
       "      <td>87.436849</td>\n",
       "      <td>9.067784</td>\n",
       "      <td>98.108210</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>6019000200</td>\n",
       "      <td>3167</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>93706</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>-119.805504</td>\n",
       "      <td>36.735491</td>\n",
       "      <td>85.970036</td>\n",
       "      <td>99.974776</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>78.389548</td>\n",
       "      <td>86.8</td>\n",
       "      <td>99.560025</td>\n",
       "      <td>16.1</td>\n",
       "      <td>87.980708</td>\n",
       "      <td>40.1</td>\n",
       "      <td>97.854785</td>\n",
       "      <td>94.581328</td>\n",
       "      <td>9.808714</td>\n",
       "      <td>99.987388</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6077000801</td>\n",
       "      <td>6692</td>\n",
       "      <td>San Joaquin</td>\n",
       "      <td>95203</td>\n",
       "      <td>Stockton</td>\n",
       "      <td>-121.314524</td>\n",
       "      <td>37.940517</td>\n",
       "      <td>82.491521</td>\n",
       "      <td>99.962164</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>75.136648</td>\n",
       "      <td>61.3</td>\n",
       "      <td>85.568825</td>\n",
       "      <td>19.6</td>\n",
       "      <td>94.973981</td>\n",
       "      <td>21.1</td>\n",
       "      <td>63.544047</td>\n",
       "      <td>86.701266</td>\n",
       "      <td>8.991499</td>\n",
       "      <td>97.717241</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6019001500</td>\n",
       "      <td>2206</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>93725</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>-119.717843</td>\n",
       "      <td>36.681600</td>\n",
       "      <td>82.030814</td>\n",
       "      <td>99.949552</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>73.723504</td>\n",
       "      <td>66.4</td>\n",
       "      <td>90.232558</td>\n",
       "      <td>18.6</td>\n",
       "      <td>93.654017</td>\n",
       "      <td>28.1</td>\n",
       "      <td>83.980706</td>\n",
       "      <td>80.075199</td>\n",
       "      <td>8.304332</td>\n",
       "      <td>92.760752</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>6037204920</td>\n",
       "      <td>2598</td>\n",
       "      <td>Los Angeles</td>\n",
       "      <td>90023</td>\n",
       "      <td>Los Angeles</td>\n",
       "      <td>-118.197497</td>\n",
       "      <td>34.017500</td>\n",
       "      <td>80.726872</td>\n",
       "      <td>99.936940</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>89.148114</td>\n",
       "      <td>66.4</td>\n",
       "      <td>90.232558</td>\n",
       "      <td>11.6</td>\n",
       "      <td>67.419723</td>\n",
       "      <td>22.0</td>\n",
       "      <td>67.034780</td>\n",
       "      <td>82.361214</td>\n",
       "      <td>8.541407</td>\n",
       "      <td>94.892168</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6077000300</td>\n",
       "      <td>2396</td>\n",
       "      <td>San Joaquin</td>\n",
       "      <td>95203</td>\n",
       "      <td>Stockton</td>\n",
       "      <td>-121.302072</td>\n",
       "      <td>37.952421</td>\n",
       "      <td>80.178935</td>\n",
       "      <td>99.924328</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>92.401013</td>\n",
       "      <td>76.2</td>\n",
       "      <td>97.033312</td>\n",
       "      <td>14.4</td>\n",
       "      <td>82.002792</td>\n",
       "      <td>24.3</td>\n",
       "      <td>74.727088</td>\n",
       "      <td>91.940147</td>\n",
       "      <td>9.534806</td>\n",
       "      <td>99.646866</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>6019001000</td>\n",
       "      <td>4106</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>93706</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>-119.804314</td>\n",
       "      <td>36.697751</td>\n",
       "      <td>80.134819</td>\n",
       "      <td>99.911716</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>76.576456</td>\n",
       "      <td>74.5</td>\n",
       "      <td>95.901948</td>\n",
       "      <td>20.0</td>\n",
       "      <td>95.494352</td>\n",
       "      <td>31.8</td>\n",
       "      <td>90.720995</td>\n",
       "      <td>93.793697</td>\n",
       "      <td>9.727032</td>\n",
       "      <td>99.911716</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>6037206050</td>\n",
       "      <td>2146</td>\n",
       "      <td>Los Angeles</td>\n",
       "      <td>90023</td>\n",
       "      <td>Los Angeles</td>\n",
       "      <td>-118.224453</td>\n",
       "      <td>34.029904</td>\n",
       "      <td>79.025293</td>\n",
       "      <td>99.899105</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>97.120384</td>\n",
       "      <td>75.7</td>\n",
       "      <td>96.643620</td>\n",
       "      <td>28.5</td>\n",
       "      <td>99.505013</td>\n",
       "      <td>31.7</td>\n",
       "      <td>90.555979</td>\n",
       "      <td>80.249569</td>\n",
       "      <td>8.322415</td>\n",
       "      <td>92.962543</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>6019000400</td>\n",
       "      <td>6343</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>93721</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>-119.776209</td>\n",
       "      <td>36.727656</td>\n",
       "      <td>78.529850</td>\n",
       "      <td>99.886493</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>71.350487</td>\n",
       "      <td>83.4</td>\n",
       "      <td>99.082338</td>\n",
       "      <td>23.5</td>\n",
       "      <td>98.273893</td>\n",
       "      <td>23.2</td>\n",
       "      <td>71.337903</td>\n",
       "      <td>83.195409</td>\n",
       "      <td>8.627918</td>\n",
       "      <td>95.711943</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10 rows × 57 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Census Tract  Total Population California County    ZIP  \\\n",
       "0    6019001100              3174           Fresno   93706   \n",
       "1    6071001600              6133    San Bernardino  91761   \n",
       "2    6019000200              3167           Fresno   93706   \n",
       "3    6077000801              6692       San Joaquin  95203   \n",
       "4    6019001500              2206           Fresno   93725   \n",
       "5    6037204920              2598       Los Angeles  90023   \n",
       "6    6077000300              2396       San Joaquin  95203   \n",
       "7    6019001000              4106           Fresno   93706   \n",
       "8    6037206050              2146       Los Angeles  90023   \n",
       "9    6019000400              6343           Fresno   93721   \n",
       "\n",
       "  Nearby City \\n(to help approximate location only)   Longitude   Latitude  \\\n",
       "0                                            Fresno -119.781696  36.709695   \n",
       "1                                           Ontario -117.618013  34.057780   \n",
       "2                                            Fresno -119.805504  36.735491   \n",
       "3                                          Stockton -121.314524  37.940517   \n",
       "4                                            Fresno -119.717843  36.681600   \n",
       "5                                       Los Angeles -118.197497  34.017500   \n",
       "6                                          Stockton -121.302072  37.952421   \n",
       "7                                            Fresno -119.804314  36.697751   \n",
       "8                                       Los Angeles -118.224453  34.029904   \n",
       "9                                            Fresno -119.776209  36.727656   \n",
       "\n",
       "   CES 3.0 Score   CES 3.0 Percentile CES 3.0 \\nPercentile Range  ...  \\\n",
       "0      94.090246           100.000000   95-100% (highest scores)  ...   \n",
       "1      90.677839            99.987388   95-100% (highest scores)  ...   \n",
       "2      85.970036            99.974776   95-100% (highest scores)  ...   \n",
       "3      82.491521            99.962164   95-100% (highest scores)  ...   \n",
       "4      82.030814            99.949552   95-100% (highest scores)  ...   \n",
       "5      80.726872            99.936940   95-100% (highest scores)  ...   \n",
       "6      80.178935            99.924328   95-100% (highest scores)  ...   \n",
       "7      80.134819            99.911716   95-100% (highest scores)  ...   \n",
       "8      79.025293            99.899105   95-100% (highest scores)  ...   \n",
       "9      78.529850            99.886493   95-100% (highest scores)  ...   \n",
       "\n",
       "  Linguistic Isolation Pctl  Poverty  Poverty Pctl  Unemployment  \\\n",
       "0                 77.509665     76.3     97.121307          17.6   \n",
       "1                 96.253833     72.5     94.632307          12.3   \n",
       "2                 78.389548     86.8     99.560025          16.1   \n",
       "3                 75.136648     61.3     85.568825          19.6   \n",
       "4                 73.723504     66.4     90.232558          18.6   \n",
       "5                 89.148114     66.4     90.232558          11.6   \n",
       "6                 92.401013     76.2     97.033312          14.4   \n",
       "7                 76.576456     74.5     95.901948          20.0   \n",
       "8                 97.120384     75.7     96.643620          28.5   \n",
       "9                 71.350487     83.4     99.082338          23.5   \n",
       "\n",
       "   Unemployment Pctl  Housing Burden  Housing Burden Pctl  Pop. Char.   \\\n",
       "0          91.724838            26.0            79.398324    92.120494   \n",
       "1          71.823836            34.1            93.754760    87.436849   \n",
       "2          87.980708            40.1            97.854785    94.581328   \n",
       "3          94.973981            21.1            63.544047    86.701266   \n",
       "4          93.654017            28.1            83.980706    80.075199   \n",
       "5          67.419723            22.0            67.034780    82.361214   \n",
       "6          82.002792            24.3            74.727088    91.940147   \n",
       "7          95.494352            31.8            90.720995    93.793697   \n",
       "8          99.505013            31.7            90.555979    80.249569   \n",
       "9          98.273893            23.2            71.337903    83.195409   \n",
       "\n",
       "   Pop. Char. Score  Pop. Char. Pctl  \n",
       "0          9.553509        99.697314  \n",
       "1          9.067784        98.108210  \n",
       "2          9.808714        99.987388  \n",
       "3          8.991499        97.717241  \n",
       "4          8.304332        92.760752  \n",
       "5          8.541407        94.892168  \n",
       "6          9.534806        99.646866  \n",
       "7          9.727032        99.911716  \n",
       "8          8.322415        92.962543  \n",
       "9          8.627918        95.711943  \n",
       "\n",
       "[10 rows x 57 columns]"
      ]
     },
     "execution_count": 172,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Select Columns and Handle Missing Values\n",
    "## Question 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "metadata": {
    "id": "krMPk6HT0KmF"
   },
   "outputs": [],
   "source": [
    "#selecting columns\n",
    "y = df['Poverty']\n",
    "X = df[['Total Population', 'Ozone', 'Ozone Pctl', 'PM2.5', 'PM2.5 Pctl', 'Diesel PM', 'Diesel PM Pctl', 'Drinking Water', 'Drinking Water Pctl', 'Pesticides', 'Pesticides Pctl', 'Tox. Release', 'Tox. Release Pctl', 'Traffic', 'Traffic Pctl', 'Cleanup Sites', 'Cleanup Sites Pctl', 'Groundwater Threats', 'Groundwater Threats Pctl', 'Haz. Waste', 'Haz. Waste Pctl', 'Imp. Water Bodies', 'Imp. Water Bodies Pctl', 'Solid Waste', 'Solid Waste Pctl', 'Asthma', 'Asthma Pctl', 'Low Birth Weight', 'Low Birth Weight Pctl', 'Cardiovascular Disease', 'Education', 'Education Pctl', 'Linguistic Isolation', 'Linguistic Isolation Pctl']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total Population               0\n",
      "Ozone                          0\n",
      "Ozone Pctl                     0\n",
      "PM2.5                         19\n",
      "PM2.5 Pctl                    19\n",
      "Diesel PM                      0\n",
      "Diesel PM Pctl                 0\n",
      "Drinking Water                18\n",
      "Drinking Water Pctl           18\n",
      "Pesticides                     0\n",
      "Pesticides Pctl                0\n",
      "Tox. Release                   0\n",
      "Tox. Release Pctl              0\n",
      "Traffic                       56\n",
      "Traffic Pctl                  56\n",
      "Cleanup Sites                  0\n",
      "Cleanup Sites Pctl             0\n",
      "Groundwater Threats            0\n",
      "Groundwater Threats Pctl       0\n",
      "Haz. Waste                     0\n",
      "Haz. Waste Pctl                0\n",
      "Imp. Water Bodies              0\n",
      "Imp. Water Bodies Pctl         0\n",
      "Solid Waste                    0\n",
      "Solid Waste Pctl               0\n",
      "Asthma                         0\n",
      "Asthma Pctl                    0\n",
      "Low Birth Weight             222\n",
      "Low Birth Weight Pctl        222\n",
      "Cardiovascular Disease         0\n",
      "Education                     96\n",
      "Education Pctl                96\n",
      "Linguistic Isolation         242\n",
      "Linguistic Isolation Pctl    242\n",
      "Poverty                       79\n",
      "dtype: int64\n"
     ]
    }
   ],
   "source": [
    "#display missing values\n",
    "missing_values = df[['Total Population', 'Ozone', 'Ozone Pctl', 'PM2.5', 'PM2.5 Pctl', 'Diesel PM', 'Diesel PM Pctl', 'Drinking Water', 'Drinking Water Pctl', 'Pesticides', 'Pesticides Pctl', 'Tox. Release', 'Tox. Release Pctl', 'Traffic', 'Traffic Pctl', 'Cleanup Sites', 'Cleanup Sites Pctl', 'Groundwater Threats', 'Groundwater Threats Pctl', 'Haz. Waste', 'Haz. Waste Pctl', 'Imp. Water Bodies', 'Imp. Water Bodies Pctl', 'Solid Waste', 'Solid Waste Pctl', 'Asthma', 'Asthma Pctl', 'Low Birth Weight', 'Low Birth Weight Pctl', 'Cardiovascular Disease', 'Education', 'Education Pctl', 'Linguistic Isolation', 'Linguistic Isolation Pctl', 'Poverty']].isnull().sum()\n",
    "print(missing_values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 7956 entries, 0 to 8032\n",
      "Data columns (total 57 columns):\n",
      " #   Column                                            Non-Null Count  Dtype  \n",
      "---  ------                                            --------------  -----  \n",
      " 0   Census Tract                                      7956 non-null   int64  \n",
      " 1   Total Population                                  7956 non-null   int64  \n",
      " 2   California County                                 7956 non-null   object \n",
      " 3   ZIP                                               7956 non-null   int64  \n",
      " 4   Nearby City \n",
      "(to help approximate location only)  7956 non-null   object \n",
      " 5   Longitude                                         7956 non-null   float64\n",
      " 6   Latitude                                          7956 non-null   float64\n",
      " 7   CES 3.0 Score                                     7926 non-null   float64\n",
      " 8    CES 3.0 Percentile                               7926 non-null   float64\n",
      " 9   CES 3.0 \n",
      "Percentile Range                         7926 non-null   object \n",
      " 10  SB 535 Disadvantaged Community                    7956 non-null   object \n",
      " 11  Ozone                                             7956 non-null   float64\n",
      " 12  Ozone Pctl                                        7956 non-null   float64\n",
      " 13  PM2.5                                             7956 non-null   float64\n",
      " 14  PM2.5 Pctl                                        7956 non-null   float64\n",
      " 15  Diesel PM                                         7956 non-null   float64\n",
      " 16  Diesel PM Pctl                                    7956 non-null   float64\n",
      " 17  Drinking Water                                    7956 non-null   float64\n",
      " 18  Drinking Water Pctl                               7956 non-null   float64\n",
      " 19  Pesticides                                        7956 non-null   float64\n",
      " 20  Pesticides Pctl                                   7956 non-null   float64\n",
      " 21  Tox. Release                                      7956 non-null   float64\n",
      " 22  Tox. Release Pctl                                 7956 non-null   float64\n",
      " 23  Traffic                                           7956 non-null   float64\n",
      " 24  Traffic Pctl                                      7956 non-null   float64\n",
      " 25  Cleanup Sites                                     7956 non-null   float64\n",
      " 26  Cleanup Sites Pctl                                7956 non-null   float64\n",
      " 27  Groundwater Threats                               7956 non-null   float64\n",
      " 28  Groundwater Threats Pctl                          7956 non-null   float64\n",
      " 29  Haz. Waste                                        7956 non-null   float64\n",
      " 30  Haz. Waste Pctl                                   7956 non-null   float64\n",
      " 31  Imp. Water Bodies                                 7956 non-null   int64  \n",
      " 32  Imp. Water Bodies Pctl                            7956 non-null   float64\n",
      " 33  Solid Waste                                       7956 non-null   float64\n",
      " 34  Solid Waste Pctl                                  7956 non-null   float64\n",
      " 35  Pollution Burden                                  7956 non-null   float64\n",
      " 36  Pollution Burden Score                            7956 non-null   float64\n",
      " 37  Pollution Burden Pctl                             7956 non-null   float64\n",
      " 38  Asthma                                            7956 non-null   float64\n",
      " 39  Asthma Pctl                                       7956 non-null   float64\n",
      " 40  Low Birth Weight                                  7956 non-null   float64\n",
      " 41  Low Birth Weight Pctl                             7956 non-null   float64\n",
      " 42  Cardiovascular Disease                            7956 non-null   float64\n",
      " 43  Cardiovascular Disease Pctl                       7956 non-null   float64\n",
      " 44  Education                                         7956 non-null   float64\n",
      " 45  Education Pctl                                    7956 non-null   float64\n",
      " 46  Linguistic Isolation                              7956 non-null   float64\n",
      " 47  Linguistic Isolation Pctl                         7956 non-null   float64\n",
      " 48  Poverty                                           7956 non-null   float64\n",
      " 49  Poverty Pctl                                      7956 non-null   float64\n",
      " 50  Unemployment                                      7870 non-null   float64\n",
      " 51  Unemployment Pctl                                 7870 non-null   float64\n",
      " 52  Housing Burden                                    7876 non-null   float64\n",
      " 53  Housing Burden Pctl                               7876 non-null   float64\n",
      " 54  Pop. Char.                                        7926 non-null   float64\n",
      " 55  Pop. Char. Score                                  7926 non-null   float64\n",
      " 56  Pop. Char. Pctl                                   7926 non-null   float64\n",
      "dtypes: float64(49), int64(4), object(4)\n",
      "memory usage: 3.5+ MB\n"
     ]
    }
   ],
   "source": [
    "#impute and fill using mean\n",
    "df['PM2.5'].fillna(df['PM2.5'].mean(), inplace=True)\n",
    "df['PM2.5 Pctl'].fillna(df['PM2.5 Pctl'].mean(), inplace=True)\n",
    "df['Drinking Water'].fillna(df['Drinking Water'].mean(), inplace=True)\n",
    "df['Drinking Water Pctl'].fillna(df['Drinking Water Pctl'].mean(), inplace=True)\n",
    "df['Traffic'].fillna(df['Traffic'].mean(), inplace=True)\n",
    "df['Traffic Pctl'].fillna(df['Traffic Pctl'].mean(), inplace=True)\n",
    "df['Low Birth Weight'].fillna(df['Low Birth Weight'].mean(), inplace=True)\n",
    "df['Low Birth Weight Pctl'].fillna(df['Low Birth Weight Pctl'].mean(), inplace=True)\n",
    "df['Education'].fillna(df['Education'].mean(), inplace=True)\n",
    "df['Education Pctl'].fillna(df['Education Pctl'].mean(), inplace=True)\n",
    "df['Linguistic Isolation'].fillna(df['Linguistic Isolation'].mean(), inplace=True)\n",
    "df['Linguistic Isolation Pctl'].fillna(df['Linguistic Isolation Pctl'].mean(), inplace=True)\n",
    "#drop NA for target variable\n",
    "df.dropna(subset=['Poverty'], inplace=True)\n",
    "\n",
    "df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I decided to drop NA's from the target variable ('Poverty'). This makes sure that the model learns from accurate data and not from imputed data points. I imputed for all feature variables that had NA's. I decided to do this because the proportion of missing data in these variables was small. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Recode Target Variable\n",
    "## Question 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Poverty mean: 36.38678984414278\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Census Tract</th>\n",
       "      <th>Total Population</th>\n",
       "      <th>California County</th>\n",
       "      <th>ZIP</th>\n",
       "      <th>Nearby City \\n(to help approximate location only)</th>\n",
       "      <th>Longitude</th>\n",
       "      <th>Latitude</th>\n",
       "      <th>CES 3.0 Score</th>\n",
       "      <th>CES 3.0 Percentile</th>\n",
       "      <th>CES 3.0 \\nPercentile Range</th>\n",
       "      <th>...</th>\n",
       "      <th>Poverty</th>\n",
       "      <th>Poverty Pctl</th>\n",
       "      <th>Unemployment</th>\n",
       "      <th>Unemployment Pctl</th>\n",
       "      <th>Housing Burden</th>\n",
       "      <th>Housing Burden Pctl</th>\n",
       "      <th>Pop. Char.</th>\n",
       "      <th>Pop. Char. Score</th>\n",
       "      <th>Pop. Char. Pctl</th>\n",
       "      <th>Poverty_recode</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6019001100</td>\n",
       "      <td>3174</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>93706</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>-119.781696</td>\n",
       "      <td>36.709695</td>\n",
       "      <td>94.090246</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>76.3</td>\n",
       "      <td>97.121307</td>\n",
       "      <td>17.6</td>\n",
       "      <td>91.724838</td>\n",
       "      <td>26.0</td>\n",
       "      <td>79.398324</td>\n",
       "      <td>92.120494</td>\n",
       "      <td>9.553509</td>\n",
       "      <td>99.697314</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6071001600</td>\n",
       "      <td>6133</td>\n",
       "      <td>San Bernardino</td>\n",
       "      <td>91761</td>\n",
       "      <td>Ontario</td>\n",
       "      <td>-117.618013</td>\n",
       "      <td>34.057780</td>\n",
       "      <td>90.677839</td>\n",
       "      <td>99.987388</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>72.5</td>\n",
       "      <td>94.632307</td>\n",
       "      <td>12.3</td>\n",
       "      <td>71.823836</td>\n",
       "      <td>34.1</td>\n",
       "      <td>93.754760</td>\n",
       "      <td>87.436849</td>\n",
       "      <td>9.067784</td>\n",
       "      <td>98.108210</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>6019000200</td>\n",
       "      <td>3167</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>93706</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>-119.805504</td>\n",
       "      <td>36.735491</td>\n",
       "      <td>85.970036</td>\n",
       "      <td>99.974776</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>86.8</td>\n",
       "      <td>99.560025</td>\n",
       "      <td>16.1</td>\n",
       "      <td>87.980708</td>\n",
       "      <td>40.1</td>\n",
       "      <td>97.854785</td>\n",
       "      <td>94.581328</td>\n",
       "      <td>9.808714</td>\n",
       "      <td>99.987388</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6077000801</td>\n",
       "      <td>6692</td>\n",
       "      <td>San Joaquin</td>\n",
       "      <td>95203</td>\n",
       "      <td>Stockton</td>\n",
       "      <td>-121.314524</td>\n",
       "      <td>37.940517</td>\n",
       "      <td>82.491521</td>\n",
       "      <td>99.962164</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>61.3</td>\n",
       "      <td>85.568825</td>\n",
       "      <td>19.6</td>\n",
       "      <td>94.973981</td>\n",
       "      <td>21.1</td>\n",
       "      <td>63.544047</td>\n",
       "      <td>86.701266</td>\n",
       "      <td>8.991499</td>\n",
       "      <td>97.717241</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6019001500</td>\n",
       "      <td>2206</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>93725</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>-119.717843</td>\n",
       "      <td>36.681600</td>\n",
       "      <td>82.030814</td>\n",
       "      <td>99.949552</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>66.4</td>\n",
       "      <td>90.232558</td>\n",
       "      <td>18.6</td>\n",
       "      <td>93.654017</td>\n",
       "      <td>28.1</td>\n",
       "      <td>83.980706</td>\n",
       "      <td>80.075199</td>\n",
       "      <td>8.304332</td>\n",
       "      <td>92.760752</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 58 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Census Tract  Total Population California County    ZIP  \\\n",
       "0    6019001100              3174           Fresno   93706   \n",
       "1    6071001600              6133    San Bernardino  91761   \n",
       "2    6019000200              3167           Fresno   93706   \n",
       "3    6077000801              6692       San Joaquin  95203   \n",
       "4    6019001500              2206           Fresno   93725   \n",
       "\n",
       "  Nearby City \\n(to help approximate location only)   Longitude   Latitude  \\\n",
       "0                                            Fresno -119.781696  36.709695   \n",
       "1                                           Ontario -117.618013  34.057780   \n",
       "2                                            Fresno -119.805504  36.735491   \n",
       "3                                          Stockton -121.314524  37.940517   \n",
       "4                                            Fresno -119.717843  36.681600   \n",
       "\n",
       "   CES 3.0 Score   CES 3.0 Percentile CES 3.0 \\nPercentile Range  ... Poverty  \\\n",
       "0      94.090246           100.000000   95-100% (highest scores)  ...    76.3   \n",
       "1      90.677839            99.987388   95-100% (highest scores)  ...    72.5   \n",
       "2      85.970036            99.974776   95-100% (highest scores)  ...    86.8   \n",
       "3      82.491521            99.962164   95-100% (highest scores)  ...    61.3   \n",
       "4      82.030814            99.949552   95-100% (highest scores)  ...    66.4   \n",
       "\n",
       "   Poverty Pctl  Unemployment  Unemployment Pctl  Housing Burden  \\\n",
       "0     97.121307          17.6          91.724838            26.0   \n",
       "1     94.632307          12.3          71.823836            34.1   \n",
       "2     99.560025          16.1          87.980708            40.1   \n",
       "3     85.568825          19.6          94.973981            21.1   \n",
       "4     90.232558          18.6          93.654017            28.1   \n",
       "\n",
       "   Housing Burden Pctl  Pop. Char.   Pop. Char. Score  Pop. Char. Pctl  \\\n",
       "0            79.398324    92.120494          9.553509        99.697314   \n",
       "1            93.754760    87.436849          9.067784        98.108210   \n",
       "2            97.854785    94.581328          9.808714        99.987388   \n",
       "3            63.544047    86.701266          8.991499        97.717241   \n",
       "4            83.980706    80.075199          8.304332        92.760752   \n",
       "\n",
       "   Poverty_recode  \n",
       "0               1  \n",
       "1               1  \n",
       "2               1  \n",
       "3               1  \n",
       "4               1  \n",
       "\n",
       "[5 rows x 58 columns]"
      ]
     },
     "execution_count": 176,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#recode 'Poverty' to 1 if greater than mean and 0 if less than mean\n",
    "df['Poverty_recode'] = np.where(df['Poverty'] <= np.mean(df['Poverty']), 0, 1)\n",
    "\n",
    "print('Poverty mean:', df['Poverty'].mean())\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Two Tables\n",
    "## Question 4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Census Tract</th>\n",
       "      <th>Total Population</th>\n",
       "      <th>California County</th>\n",
       "      <th>ZIP</th>\n",
       "      <th>Nearby City \\n(to help approximate location only)</th>\n",
       "      <th>Longitude</th>\n",
       "      <th>Latitude</th>\n",
       "      <th>CES 3.0 Score</th>\n",
       "      <th>CES 3.0 Percentile</th>\n",
       "      <th>CES 3.0 \\nPercentile Range</th>\n",
       "      <th>...</th>\n",
       "      <th>Poverty</th>\n",
       "      <th>Poverty Pctl</th>\n",
       "      <th>Unemployment</th>\n",
       "      <th>Unemployment Pctl</th>\n",
       "      <th>Housing Burden</th>\n",
       "      <th>Housing Burden Pctl</th>\n",
       "      <th>Pop. Char.</th>\n",
       "      <th>Pop. Char. Score</th>\n",
       "      <th>Pop. Char. Pctl</th>\n",
       "      <th>Poverty_recode</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>6037543305</td>\n",
       "      <td>2666</td>\n",
       "      <td>Los Angeles</td>\n",
       "      <td>90220</td>\n",
       "      <td>Compton</td>\n",
       "      <td>-118.223869</td>\n",
       "      <td>33.861147</td>\n",
       "      <td>73.535593</td>\n",
       "      <td>99.596418</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>34.1</td>\n",
       "      <td>50.949089</td>\n",
       "      <td>12.1</td>\n",
       "      <td>70.567331</td>\n",
       "      <td>30.6</td>\n",
       "      <td>88.575781</td>\n",
       "      <td>75.906646</td>\n",
       "      <td>7.872025</td>\n",
       "      <td>88.258292</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>6077005119</td>\n",
       "      <td>6006</td>\n",
       "      <td>San Joaquin</td>\n",
       "      <td>95330</td>\n",
       "      <td>Lathrop</td>\n",
       "      <td>-121.302745</td>\n",
       "      <td>37.844684</td>\n",
       "      <td>71.299913</td>\n",
       "      <td>99.457687</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>34.7</td>\n",
       "      <td>52.093023</td>\n",
       "      <td>11.3</td>\n",
       "      <td>65.554004</td>\n",
       "      <td>18.8</td>\n",
       "      <td>53.706524</td>\n",
       "      <td>69.682413</td>\n",
       "      <td>7.226531</td>\n",
       "      <td>79.467777</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>151</th>\n",
       "      <td>6037543100</td>\n",
       "      <td>6759</td>\n",
       "      <td>Los Angeles</td>\n",
       "      <td>90220</td>\n",
       "      <td>Compton</td>\n",
       "      <td>-118.253953</td>\n",
       "      <td>33.881033</td>\n",
       "      <td>64.637236</td>\n",
       "      <td>98.095598</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>35.5</td>\n",
       "      <td>53.161534</td>\n",
       "      <td>11.7</td>\n",
       "      <td>68.168549</td>\n",
       "      <td>26.6</td>\n",
       "      <td>80.807312</td>\n",
       "      <td>74.894195</td>\n",
       "      <td>7.767027</td>\n",
       "      <td>86.845756</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>215</th>\n",
       "      <td>6037532303</td>\n",
       "      <td>4464</td>\n",
       "      <td>Los Angeles</td>\n",
       "      <td>90040</td>\n",
       "      <td>Los Angeles</td>\n",
       "      <td>-118.146343</td>\n",
       "      <td>34.000329</td>\n",
       "      <td>62.344937</td>\n",
       "      <td>97.288435</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>33.2</td>\n",
       "      <td>49.654305</td>\n",
       "      <td>18.2</td>\n",
       "      <td>92.930575</td>\n",
       "      <td>19.9</td>\n",
       "      <td>58.644326</td>\n",
       "      <td>65.260421</td>\n",
       "      <td>6.767941</td>\n",
       "      <td>73.023080</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>223</th>\n",
       "      <td>6037541002</td>\n",
       "      <td>3209</td>\n",
       "      <td>Los Angeles</td>\n",
       "      <td>90746</td>\n",
       "      <td>Carson</td>\n",
       "      <td>-118.272812</td>\n",
       "      <td>33.875181</td>\n",
       "      <td>62.201483</td>\n",
       "      <td>97.187539</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>35.9</td>\n",
       "      <td>53.777498</td>\n",
       "      <td>16.9</td>\n",
       "      <td>90.201802</td>\n",
       "      <td>21.0</td>\n",
       "      <td>63.201320</td>\n",
       "      <td>63.520645</td>\n",
       "      <td>6.587514</td>\n",
       "      <td>70.324127</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 58 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     Census Tract  Total Population California County    ZIP  \\\n",
       "32     6037543305              2666       Los Angeles  90220   \n",
       "43     6077005119              6006       San Joaquin  95330   \n",
       "151    6037543100              6759       Los Angeles  90220   \n",
       "215    6037532303              4464       Los Angeles  90040   \n",
       "223    6037541002              3209       Los Angeles  90746   \n",
       "\n",
       "    Nearby City \\n(to help approximate location only)   Longitude   Latitude  \\\n",
       "32                                            Compton -118.223869  33.861147   \n",
       "43                                            Lathrop -121.302745  37.844684   \n",
       "151                                           Compton -118.253953  33.881033   \n",
       "215                                       Los Angeles -118.146343  34.000329   \n",
       "223                                            Carson -118.272812  33.875181   \n",
       "\n",
       "     CES 3.0 Score   CES 3.0 Percentile CES 3.0 \\nPercentile Range  ...  \\\n",
       "32       73.535593            99.596418   95-100% (highest scores)  ...   \n",
       "43       71.299913            99.457687   95-100% (highest scores)  ...   \n",
       "151      64.637236            98.095598   95-100% (highest scores)  ...   \n",
       "215      62.344937            97.288435   95-100% (highest scores)  ...   \n",
       "223      62.201483            97.187539   95-100% (highest scores)  ...   \n",
       "\n",
       "    Poverty  Poverty Pctl  Unemployment  Unemployment Pctl  Housing Burden  \\\n",
       "32     34.1     50.949089          12.1          70.567331            30.6   \n",
       "43     34.7     52.093023          11.3          65.554004            18.8   \n",
       "151    35.5     53.161534          11.7          68.168549            26.6   \n",
       "215    33.2     49.654305          18.2          92.930575            19.9   \n",
       "223    35.9     53.777498          16.9          90.201802            21.0   \n",
       "\n",
       "     Housing Burden Pctl  Pop. Char.   Pop. Char. Score  Pop. Char. Pctl  \\\n",
       "32             88.575781    75.906646          7.872025        88.258292   \n",
       "43             53.706524    69.682413          7.226531        79.467777   \n",
       "151            80.807312    74.894195          7.767027        86.845756   \n",
       "215            58.644326    65.260421          6.767941        73.023080   \n",
       "223            63.201320    63.520645          6.587514        70.324127   \n",
       "\n",
       "     Poverty_recode  \n",
       "32                0  \n",
       "43                0  \n",
       "151               0  \n",
       "215               0  \n",
       "223               0  \n",
       "\n",
       "[5 rows x 58 columns]"
      ]
     },
     "execution_count": 177,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#subset rows where 'Poverty_recode' = 0, called 'lowData'\n",
    "lowData = df[df['Poverty_recode'] == 0]\n",
    "                                           \n",
    "lowData.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This table only shows data where the column 'Poverty_recode' is equal to 0. This means all the data in this table has a poverty value below the mean (36.38678984414278). I made this table to show how to subset the data in case you only wanted to analyze the 'lowData'. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Census Tract</th>\n",
       "      <th>Total Population</th>\n",
       "      <th>California County</th>\n",
       "      <th>ZIP</th>\n",
       "      <th>Nearby City \\n(to help approximate location only)</th>\n",
       "      <th>Longitude</th>\n",
       "      <th>Latitude</th>\n",
       "      <th>CES 3.0 Score</th>\n",
       "      <th>CES 3.0 Percentile</th>\n",
       "      <th>CES 3.0 \\nPercentile Range</th>\n",
       "      <th>...</th>\n",
       "      <th>Poverty</th>\n",
       "      <th>Poverty Pctl</th>\n",
       "      <th>Unemployment</th>\n",
       "      <th>Unemployment Pctl</th>\n",
       "      <th>Housing Burden</th>\n",
       "      <th>Housing Burden Pctl</th>\n",
       "      <th>Pop. Char.</th>\n",
       "      <th>Pop. Char. Score</th>\n",
       "      <th>Pop. Char. Pctl</th>\n",
       "      <th>Poverty_recode</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6019001100</td>\n",
       "      <td>3174</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>93706</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>-119.781696</td>\n",
       "      <td>36.709695</td>\n",
       "      <td>94.090246</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>76.3</td>\n",
       "      <td>97.121307</td>\n",
       "      <td>17.6</td>\n",
       "      <td>91.724838</td>\n",
       "      <td>26.0</td>\n",
       "      <td>79.398324</td>\n",
       "      <td>92.120494</td>\n",
       "      <td>9.553509</td>\n",
       "      <td>99.697314</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6071001600</td>\n",
       "      <td>6133</td>\n",
       "      <td>San Bernardino</td>\n",
       "      <td>91761</td>\n",
       "      <td>Ontario</td>\n",
       "      <td>-117.618013</td>\n",
       "      <td>34.057780</td>\n",
       "      <td>90.677839</td>\n",
       "      <td>99.987388</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>72.5</td>\n",
       "      <td>94.632307</td>\n",
       "      <td>12.3</td>\n",
       "      <td>71.823836</td>\n",
       "      <td>34.1</td>\n",
       "      <td>93.754760</td>\n",
       "      <td>87.436849</td>\n",
       "      <td>9.067784</td>\n",
       "      <td>98.108210</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>6019000200</td>\n",
       "      <td>3167</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>93706</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>-119.805504</td>\n",
       "      <td>36.735491</td>\n",
       "      <td>85.970036</td>\n",
       "      <td>99.974776</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>86.8</td>\n",
       "      <td>99.560025</td>\n",
       "      <td>16.1</td>\n",
       "      <td>87.980708</td>\n",
       "      <td>40.1</td>\n",
       "      <td>97.854785</td>\n",
       "      <td>94.581328</td>\n",
       "      <td>9.808714</td>\n",
       "      <td>99.987388</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6077000801</td>\n",
       "      <td>6692</td>\n",
       "      <td>San Joaquin</td>\n",
       "      <td>95203</td>\n",
       "      <td>Stockton</td>\n",
       "      <td>-121.314524</td>\n",
       "      <td>37.940517</td>\n",
       "      <td>82.491521</td>\n",
       "      <td>99.962164</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>61.3</td>\n",
       "      <td>85.568825</td>\n",
       "      <td>19.6</td>\n",
       "      <td>94.973981</td>\n",
       "      <td>21.1</td>\n",
       "      <td>63.544047</td>\n",
       "      <td>86.701266</td>\n",
       "      <td>8.991499</td>\n",
       "      <td>97.717241</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6019001500</td>\n",
       "      <td>2206</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>93725</td>\n",
       "      <td>Fresno</td>\n",
       "      <td>-119.717843</td>\n",
       "      <td>36.681600</td>\n",
       "      <td>82.030814</td>\n",
       "      <td>99.949552</td>\n",
       "      <td>95-100% (highest scores)</td>\n",
       "      <td>...</td>\n",
       "      <td>66.4</td>\n",
       "      <td>90.232558</td>\n",
       "      <td>18.6</td>\n",
       "      <td>93.654017</td>\n",
       "      <td>28.1</td>\n",
       "      <td>83.980706</td>\n",
       "      <td>80.075199</td>\n",
       "      <td>8.304332</td>\n",
       "      <td>92.760752</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 58 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Census Tract  Total Population California County    ZIP  \\\n",
       "0    6019001100              3174           Fresno   93706   \n",
       "1    6071001600              6133    San Bernardino  91761   \n",
       "2    6019000200              3167           Fresno   93706   \n",
       "3    6077000801              6692       San Joaquin  95203   \n",
       "4    6019001500              2206           Fresno   93725   \n",
       "\n",
       "  Nearby City \\n(to help approximate location only)   Longitude   Latitude  \\\n",
       "0                                            Fresno -119.781696  36.709695   \n",
       "1                                           Ontario -117.618013  34.057780   \n",
       "2                                            Fresno -119.805504  36.735491   \n",
       "3                                          Stockton -121.314524  37.940517   \n",
       "4                                            Fresno -119.717843  36.681600   \n",
       "\n",
       "   CES 3.0 Score   CES 3.0 Percentile CES 3.0 \\nPercentile Range  ... Poverty  \\\n",
       "0      94.090246           100.000000   95-100% (highest scores)  ...    76.3   \n",
       "1      90.677839            99.987388   95-100% (highest scores)  ...    72.5   \n",
       "2      85.970036            99.974776   95-100% (highest scores)  ...    86.8   \n",
       "3      82.491521            99.962164   95-100% (highest scores)  ...    61.3   \n",
       "4      82.030814            99.949552   95-100% (highest scores)  ...    66.4   \n",
       "\n",
       "   Poverty Pctl  Unemployment  Unemployment Pctl  Housing Burden  \\\n",
       "0     97.121307          17.6          91.724838            26.0   \n",
       "1     94.632307          12.3          71.823836            34.1   \n",
       "2     99.560025          16.1          87.980708            40.1   \n",
       "3     85.568825          19.6          94.973981            21.1   \n",
       "4     90.232558          18.6          93.654017            28.1   \n",
       "\n",
       "   Housing Burden Pctl  Pop. Char.   Pop. Char. Score  Pop. Char. Pctl  \\\n",
       "0            79.398324    92.120494          9.553509        99.697314   \n",
       "1            93.754760    87.436849          9.067784        98.108210   \n",
       "2            97.854785    94.581328          9.808714        99.987388   \n",
       "3            63.544047    86.701266          8.991499        97.717241   \n",
       "4            83.980706    80.075199          8.304332        92.760752   \n",
       "\n",
       "   Poverty_recode  \n",
       "0               1  \n",
       "1               1  \n",
       "2               1  \n",
       "3               1  \n",
       "4               1  \n",
       "\n",
       "[5 rows x 58 columns]"
      ]
     },
     "execution_count": 178,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#subset rows where 'Poverty_recode' = 1, called 'highData'\n",
    "highData = df[df['Poverty_recode'] == 1]\n",
    "                                            \n",
    "highData.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This table only shows data where the column 'Poverty_recode' is equal to 1. This means all the data in this table has a poverty value above the mean (36.38678984414278). I made this table to show how to subset the data in case you only wanted to analyze the 'highData'. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(7956, 34) (7956,)\n"
     ]
    }
   ],
   "source": [
    "#dropping 'Poverty' as target variable, adding 'Poverty_recode' as target variable\n",
    "y = df.drop('Poverty', axis=1)\n",
    "y = df['Poverty_recode']\n",
    "X = df[['Total Population', 'Ozone', 'Ozone Pctl', 'PM2.5', 'PM2.5 Pctl', 'Diesel PM', 'Diesel PM Pctl', 'Drinking Water', 'Drinking Water Pctl', 'Pesticides', 'Pesticides Pctl', 'Tox. Release', 'Tox. Release Pctl', 'Traffic', 'Traffic Pctl', 'Cleanup Sites', 'Cleanup Sites Pctl', 'Groundwater Threats', 'Groundwater Threats Pctl', 'Haz. Waste', 'Haz. Waste Pctl', 'Imp. Water Bodies', 'Imp. Water Bodies Pctl', 'Solid Waste', 'Solid Waste Pctl', 'Asthma', 'Asthma Pctl', 'Low Birth Weight', 'Low Birth Weight Pctl', 'Cardiovascular Disease', 'Education', 'Education Pctl', 'Linguistic Isolation', 'Linguistic Isolation Pctl']]\n",
    "print(X.shape, y.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Split data into X and y\n",
    "# StandardScalar\n",
    "## Question 5 and 6"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {},
   "outputs": [],
   "source": [
    "#convert to numpy arrays\n",
    "X = np.array(X)\n",
    "\n",
    "#90/10 split using student ID random_state\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.1, random_state=2735566)\n",
    "\n",
    "#initialize StandardScaler\n",
    "scaler = StandardScaler()\n",
    "\n",
    "#fit scaler on training data\n",
    "X_train_scaled = scaler.fit_transform(X_train)\n",
    "\n",
    "#fit scaler on test data\n",
    "X_test_scaled = scaler.transform(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Build model\n",
    "## Question 7"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 304,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_26\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " dense_81 (Dense)            (None, 34)                1190      \n",
      "                                                                 \n",
      " dropout_53 (Dropout)        (None, 34)                0         \n",
      "                                                                 \n",
      " dense_82 (Dense)            (None, 34)                1190      \n",
      "                                                                 \n",
      " dropout_54 (Dropout)        (None, 34)                0         \n",
      "                                                                 \n",
      " dense_83 (Dense)            (None, 1)                 35        \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 2415 (9.43 KB)\n",
      "Trainable params: 2415 (9.43 KB)\n",
      "Non-trainable params: 0 (0.00 Byte)\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "#initialize model\n",
    "model = Sequential()\n",
    "\n",
    "#add input layer, specify shape of input data, 'relu' activation function\n",
    "model.add(Dense(34, input_shape=(X.shape[1],), activation='relu'))\n",
    "\n",
    "#add dropout layer\n",
    "model.add(Dropout(0.1))\n",
    "\n",
    "model.add(Dense(34, activation='relu'))\n",
    "model.add(Dropout(0.1))\n",
    "\n",
    "#add output layer, 'sigmoid' activation function\n",
    "model.add(Dense(1, activation='sigmoid'))\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 305,
   "metadata": {},
   "outputs": [],
   "source": [
    "#compiling NN model\n",
    "model.compile(optimizer='adam', #'adam' optimizer\n",
    "              loss='binary_crossentropy', #binary crossentropy for binary classification\n",
    "              metrics=['accuracy']) #accuracy is metric for evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 306,
   "metadata": {},
   "outputs": [],
   "source": [
    "#initialize EarlyStopping callback\n",
    "es = EarlyStopping(monitor='val_loss', patience=20, restore_best_weights=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Fit Model\n",
    "## Question 8"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 307,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100000\n",
      "144/144 [==============================] - 1s 3ms/step - loss: 77.6380 - accuracy: 0.5237 - val_loss: 16.8527 - val_accuracy: 0.5817\n",
      "Epoch 2/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 30.6430 - accuracy: 0.5473 - val_loss: 7.0337 - val_accuracy: 0.5955\n",
      "Epoch 3/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 15.3409 - accuracy: 0.5631 - val_loss: 2.6655 - val_accuracy: 0.6281\n",
      "Epoch 4/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 10.8843 - accuracy: 0.5647 - val_loss: 1.5635 - val_accuracy: 0.6357\n",
      "Epoch 5/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 6.4282 - accuracy: 0.5539 - val_loss: 0.9637 - val_accuracy: 0.5641\n",
      "Epoch 6/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 3.9862 - accuracy: 0.5536 - val_loss: 0.8250 - val_accuracy: 0.5641\n",
      "Epoch 7/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 2.5265 - accuracy: 0.5444 - val_loss: 1.3820 - val_accuracy: 0.5641\n",
      "Epoch 8/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 2.5988 - accuracy: 0.5363 - val_loss: 0.8932 - val_accuracy: 0.5553\n",
      "Epoch 9/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 2.1342 - accuracy: 0.5388 - val_loss: 0.9027 - val_accuracy: 0.5515\n",
      "Epoch 10/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 1.7341 - accuracy: 0.5392 - val_loss: 0.7294 - val_accuracy: 0.5490\n",
      "Epoch 11/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 1.4433 - accuracy: 0.5440 - val_loss: 0.6769 - val_accuracy: 0.5515\n",
      "Epoch 12/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 1.4638 - accuracy: 0.5331 - val_loss: 0.6927 - val_accuracy: 0.5540\n",
      "Epoch 13/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.9931 - accuracy: 0.5355 - val_loss: 0.6910 - val_accuracy: 0.5678\n",
      "Epoch 14/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 1.0468 - accuracy: 0.5373 - val_loss: 0.6859 - val_accuracy: 0.5528\n",
      "Epoch 15/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 1.0686 - accuracy: 0.5415 - val_loss: 0.6917 - val_accuracy: 0.5515\n",
      "Epoch 16/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.8320 - accuracy: 0.5482 - val_loss: 0.6853 - val_accuracy: 0.5452\n",
      "Epoch 17/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.8838 - accuracy: 0.5472 - val_loss: 0.6862 - val_accuracy: 0.5415\n",
      "Epoch 18/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.8042 - accuracy: 0.5503 - val_loss: 0.6824 - val_accuracy: 0.5653\n",
      "Epoch 19/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.8298 - accuracy: 0.5513 - val_loss: 0.6867 - val_accuracy: 0.5440\n",
      "Epoch 20/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.7592 - accuracy: 0.5503 - val_loss: 0.6828 - val_accuracy: 0.5503\n",
      "Epoch 21/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.7403 - accuracy: 0.5663 - val_loss: 0.6818 - val_accuracy: 0.5704\n",
      "Epoch 22/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.8215 - accuracy: 0.5601 - val_loss: 0.6778 - val_accuracy: 0.5779\n",
      "Epoch 23/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.7135 - accuracy: 0.5729 - val_loss: 0.6774 - val_accuracy: 0.5879\n",
      "Epoch 24/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.7150 - accuracy: 0.5764 - val_loss: 0.6634 - val_accuracy: 0.6106\n",
      "Epoch 25/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.7164 - accuracy: 0.5966 - val_loss: 0.6411 - val_accuracy: 0.6482\n",
      "Epoch 26/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.6848 - accuracy: 0.6112 - val_loss: 0.6434 - val_accuracy: 0.6369\n",
      "Epoch 27/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.6823 - accuracy: 0.6228 - val_loss: 0.6324 - val_accuracy: 0.6495\n",
      "Epoch 28/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.6776 - accuracy: 0.6215 - val_loss: 0.6445 - val_accuracy: 0.6495\n",
      "Epoch 29/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.6923 - accuracy: 0.6355 - val_loss: 0.6527 - val_accuracy: 0.6244\n",
      "Epoch 30/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.6568 - accuracy: 0.6508 - val_loss: 0.6156 - val_accuracy: 0.7286\n",
      "Epoch 31/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.6500 - accuracy: 0.6638 - val_loss: 0.6033 - val_accuracy: 0.7035\n",
      "Epoch 32/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.6437 - accuracy: 0.6655 - val_loss: 0.6225 - val_accuracy: 0.6570\n",
      "Epoch 33/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.6282 - accuracy: 0.6816 - val_loss: 0.5763 - val_accuracy: 0.7136\n",
      "Epoch 34/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.6249 - accuracy: 0.6789 - val_loss: 0.5615 - val_accuracy: 0.7500\n",
      "Epoch 35/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.6121 - accuracy: 0.6902 - val_loss: 0.6104 - val_accuracy: 0.6746\n",
      "Epoch 36/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.6037 - accuracy: 0.6955 - val_loss: 0.5429 - val_accuracy: 0.7588\n",
      "Epoch 37/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.6006 - accuracy: 0.7054 - val_loss: 0.5432 - val_accuracy: 0.7475\n",
      "Epoch 38/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.5839 - accuracy: 0.7088 - val_loss: 0.5463 - val_accuracy: 0.7638\n",
      "Epoch 39/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.5712 - accuracy: 0.7190 - val_loss: 0.5309 - val_accuracy: 0.7588\n",
      "Epoch 40/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.5874 - accuracy: 0.7182 - val_loss: 0.5183 - val_accuracy: 0.7852\n",
      "Epoch 41/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.5525 - accuracy: 0.7313 - val_loss: 0.5034 - val_accuracy: 0.7802\n",
      "Epoch 42/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.5525 - accuracy: 0.7355 - val_loss: 0.4851 - val_accuracy: 0.8028\n",
      "Epoch 43/100000\n",
      "144/144 [==============================] - 1s 4ms/step - loss: 0.5705 - accuracy: 0.7395 - val_loss: 0.4873 - val_accuracy: 0.7965\n",
      "Epoch 44/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.5389 - accuracy: 0.7418 - val_loss: 0.4583 - val_accuracy: 0.8028\n",
      "Epoch 45/100000\n",
      "144/144 [==============================] - 1s 6ms/step - loss: 0.5364 - accuracy: 0.7472 - val_loss: 0.4840 - val_accuracy: 0.7915\n",
      "Epoch 46/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.5377 - accuracy: 0.7482 - val_loss: 0.4705 - val_accuracy: 0.8015\n",
      "Epoch 47/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.5126 - accuracy: 0.7556 - val_loss: 0.4622 - val_accuracy: 0.8003\n",
      "Epoch 48/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.5066 - accuracy: 0.7605 - val_loss: 0.4518 - val_accuracy: 0.8191\n",
      "Epoch 49/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.4958 - accuracy: 0.7663 - val_loss: 0.4262 - val_accuracy: 0.8191\n",
      "Epoch 50/100000\n",
      "144/144 [==============================] - 1s 4ms/step - loss: 0.4902 - accuracy: 0.7736 - val_loss: 0.4089 - val_accuracy: 0.8317\n",
      "Epoch 51/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4881 - accuracy: 0.7750 - val_loss: 0.4231 - val_accuracy: 0.8178\n",
      "Epoch 52/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4826 - accuracy: 0.7764 - val_loss: 0.4492 - val_accuracy: 0.8116\n",
      "Epoch 53/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4724 - accuracy: 0.7821 - val_loss: 0.4167 - val_accuracy: 0.8216\n",
      "Epoch 54/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4663 - accuracy: 0.7810 - val_loss: 0.4375 - val_accuracy: 0.8204\n",
      "Epoch 55/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4612 - accuracy: 0.7848 - val_loss: 0.4363 - val_accuracy: 0.8204\n",
      "Epoch 56/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4542 - accuracy: 0.7865 - val_loss: 0.4136 - val_accuracy: 0.8178\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 57/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4541 - accuracy: 0.7909 - val_loss: 0.3912 - val_accuracy: 0.8317\n",
      "Epoch 58/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4515 - accuracy: 0.7862 - val_loss: 0.3911 - val_accuracy: 0.8291\n",
      "Epoch 59/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4426 - accuracy: 0.7940 - val_loss: 0.3897 - val_accuracy: 0.8329\n",
      "Epoch 60/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4403 - accuracy: 0.7940 - val_loss: 0.4145 - val_accuracy: 0.8304\n",
      "Epoch 61/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4401 - accuracy: 0.7962 - val_loss: 0.3770 - val_accuracy: 0.8405\n",
      "Epoch 62/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4323 - accuracy: 0.7976 - val_loss: 0.3806 - val_accuracy: 0.8354\n",
      "Epoch 63/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4285 - accuracy: 0.8056 - val_loss: 0.4080 - val_accuracy: 0.8279\n",
      "Epoch 64/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4273 - accuracy: 0.8050 - val_loss: 0.3793 - val_accuracy: 0.8392\n",
      "Epoch 65/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4183 - accuracy: 0.8032 - val_loss: 0.3636 - val_accuracy: 0.8417\n",
      "Epoch 66/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4308 - accuracy: 0.8027 - val_loss: 0.3802 - val_accuracy: 0.8317\n",
      "Epoch 67/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.6007 - accuracy: 0.7962 - val_loss: 0.3786 - val_accuracy: 0.8367\n",
      "Epoch 68/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4384 - accuracy: 0.8017 - val_loss: 0.3850 - val_accuracy: 0.8266\n",
      "Epoch 69/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4275 - accuracy: 0.8085 - val_loss: 0.3945 - val_accuracy: 0.8266\n",
      "Epoch 70/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4160 - accuracy: 0.8071 - val_loss: 0.3591 - val_accuracy: 0.8342\n",
      "Epoch 71/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4116 - accuracy: 0.8144 - val_loss: 0.3554 - val_accuracy: 0.8405\n",
      "Epoch 72/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4214 - accuracy: 0.8126 - val_loss: 0.3582 - val_accuracy: 0.8392\n",
      "Epoch 73/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4106 - accuracy: 0.8154 - val_loss: 0.3638 - val_accuracy: 0.8279\n",
      "Epoch 74/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4141 - accuracy: 0.8145 - val_loss: 0.3765 - val_accuracy: 0.8241\n",
      "Epoch 75/100000\n",
      "144/144 [==============================] - 1s 4ms/step - loss: 0.4233 - accuracy: 0.8169 - val_loss: 0.3556 - val_accuracy: 0.8430\n",
      "Epoch 76/100000\n",
      "144/144 [==============================] - 1s 4ms/step - loss: 0.4075 - accuracy: 0.8148 - val_loss: 0.3667 - val_accuracy: 0.8291\n",
      "Epoch 77/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4058 - accuracy: 0.8186 - val_loss: 0.3454 - val_accuracy: 0.8392\n",
      "Epoch 78/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.4022 - accuracy: 0.8237 - val_loss: 0.3674 - val_accuracy: 0.8304\n",
      "Epoch 79/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3905 - accuracy: 0.8293 - val_loss: 0.3464 - val_accuracy: 0.8518\n",
      "Epoch 80/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3971 - accuracy: 0.8242 - val_loss: 0.3551 - val_accuracy: 0.8417\n",
      "Epoch 81/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.4142 - accuracy: 0.8306 - val_loss: 0.3560 - val_accuracy: 0.8379\n",
      "Epoch 82/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3915 - accuracy: 0.8274 - val_loss: 0.3424 - val_accuracy: 0.8467\n",
      "Epoch 83/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3869 - accuracy: 0.8284 - val_loss: 0.3384 - val_accuracy: 0.8555\n",
      "Epoch 84/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3950 - accuracy: 0.8291 - val_loss: 0.3497 - val_accuracy: 0.8442\n",
      "Epoch 85/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3863 - accuracy: 0.8278 - val_loss: 0.3493 - val_accuracy: 0.8367\n",
      "Epoch 86/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.3811 - accuracy: 0.8311 - val_loss: 0.3510 - val_accuracy: 0.8317\n",
      "Epoch 87/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3840 - accuracy: 0.8325 - val_loss: 0.3416 - val_accuracy: 0.8442\n",
      "Epoch 88/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3859 - accuracy: 0.8321 - val_loss: 0.3309 - val_accuracy: 0.8505\n",
      "Epoch 89/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3879 - accuracy: 0.8314 - val_loss: 0.3392 - val_accuracy: 0.8492\n",
      "Epoch 90/100000\n",
      "144/144 [==============================] - 1s 4ms/step - loss: 0.3829 - accuracy: 0.8345 - val_loss: 0.3302 - val_accuracy: 0.8518\n",
      "Epoch 91/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3887 - accuracy: 0.8306 - val_loss: 0.3403 - val_accuracy: 0.8518\n",
      "Epoch 92/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3805 - accuracy: 0.8323 - val_loss: 0.3624 - val_accuracy: 0.8367\n",
      "Epoch 93/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3899 - accuracy: 0.8284 - val_loss: 0.3404 - val_accuracy: 0.8467\n",
      "Epoch 94/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3779 - accuracy: 0.8349 - val_loss: 0.3329 - val_accuracy: 0.8480\n",
      "Epoch 95/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3802 - accuracy: 0.8331 - val_loss: 0.3280 - val_accuracy: 0.8606\n",
      "Epoch 96/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3747 - accuracy: 0.8341 - val_loss: 0.3342 - val_accuracy: 0.8480\n",
      "Epoch 97/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3827 - accuracy: 0.8342 - val_loss: 0.3217 - val_accuracy: 0.8518\n",
      "Epoch 98/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3782 - accuracy: 0.8356 - val_loss: 0.3314 - val_accuracy: 0.8480\n",
      "Epoch 99/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3763 - accuracy: 0.8355 - val_loss: 0.3228 - val_accuracy: 0.8455\n",
      "Epoch 100/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3796 - accuracy: 0.8358 - val_loss: 0.3284 - val_accuracy: 0.8505\n",
      "Epoch 101/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3916 - accuracy: 0.8383 - val_loss: 0.3523 - val_accuracy: 0.8455\n",
      "Epoch 102/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3741 - accuracy: 0.8385 - val_loss: 0.3289 - val_accuracy: 0.8555\n",
      "Epoch 103/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3708 - accuracy: 0.8412 - val_loss: 0.3366 - val_accuracy: 0.8405\n",
      "Epoch 104/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3757 - accuracy: 0.8359 - val_loss: 0.3351 - val_accuracy: 0.8430\n",
      "Epoch 105/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3730 - accuracy: 0.8408 - val_loss: 0.3275 - val_accuracy: 0.8593\n",
      "Epoch 106/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3703 - accuracy: 0.8374 - val_loss: 0.3403 - val_accuracy: 0.8455\n",
      "Epoch 107/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3851 - accuracy: 0.8316 - val_loss: 0.3611 - val_accuracy: 0.8229\n",
      "Epoch 108/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3711 - accuracy: 0.8394 - val_loss: 0.3367 - val_accuracy: 0.8518\n",
      "Epoch 109/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3709 - accuracy: 0.8423 - val_loss: 0.3323 - val_accuracy: 0.8480\n",
      "Epoch 110/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3742 - accuracy: 0.8366 - val_loss: 0.3400 - val_accuracy: 0.8392\n",
      "Epoch 111/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3777 - accuracy: 0.8332 - val_loss: 0.3508 - val_accuracy: 0.8304\n",
      "Epoch 112/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3687 - accuracy: 0.8374 - val_loss: 0.3287 - val_accuracy: 0.8455\n",
      "Epoch 113/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3695 - accuracy: 0.8429 - val_loss: 0.3258 - val_accuracy: 0.8505\n",
      "Epoch 114/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3649 - accuracy: 0.8436 - val_loss: 0.3239 - val_accuracy: 0.8555\n",
      "Epoch 115/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3706 - accuracy: 0.8425 - val_loss: 0.3112 - val_accuracy: 0.8505\n",
      "Epoch 116/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3600 - accuracy: 0.8440 - val_loss: 0.3227 - val_accuracy: 0.8530\n",
      "Epoch 117/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3635 - accuracy: 0.8399 - val_loss: 0.3195 - val_accuracy: 0.8543\n",
      "Epoch 118/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3583 - accuracy: 0.8465 - val_loss: 0.3176 - val_accuracy: 0.8555\n",
      "Epoch 119/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3580 - accuracy: 0.8425 - val_loss: 0.3224 - val_accuracy: 0.8568\n",
      "Epoch 120/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3645 - accuracy: 0.8446 - val_loss: 0.3232 - val_accuracy: 0.8442\n",
      "Epoch 121/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.3556 - accuracy: 0.8459 - val_loss: 0.3260 - val_accuracy: 0.8555\n",
      "Epoch 122/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3567 - accuracy: 0.8464 - val_loss: 0.3614 - val_accuracy: 0.8317\n",
      "Epoch 123/100000\n",
      "144/144 [==============================] - 1s 4ms/step - loss: 0.3555 - accuracy: 0.8462 - val_loss: 0.3132 - val_accuracy: 0.8543\n",
      "Epoch 124/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.4012 - accuracy: 0.8483 - val_loss: 0.3185 - val_accuracy: 0.8606\n",
      "Epoch 125/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3518 - accuracy: 0.8500 - val_loss: 0.3253 - val_accuracy: 0.8518\n",
      "Epoch 126/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.3572 - accuracy: 0.8472 - val_loss: 0.3180 - val_accuracy: 0.8505\n",
      "Epoch 127/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3539 - accuracy: 0.8496 - val_loss: 0.3788 - val_accuracy: 0.8204\n",
      "Epoch 128/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3518 - accuracy: 0.8485 - val_loss: 0.3245 - val_accuracy: 0.8568\n",
      "Epoch 129/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3485 - accuracy: 0.8525 - val_loss: 0.3154 - val_accuracy: 0.8706\n",
      "Epoch 130/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3563 - accuracy: 0.8487 - val_loss: 0.3305 - val_accuracy: 0.8455\n",
      "Epoch 131/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.3534 - accuracy: 0.8499 - val_loss: 0.3289 - val_accuracy: 0.8518\n",
      "Epoch 132/100000\n",
      "144/144 [==============================] - 1s 3ms/step - loss: 0.3507 - accuracy: 0.8475 - val_loss: 0.3193 - val_accuracy: 0.8568\n",
      "Epoch 133/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3543 - accuracy: 0.8471 - val_loss: 0.3105 - val_accuracy: 0.8568\n",
      "Epoch 134/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.3492 - accuracy: 0.8510 - val_loss: 0.3084 - val_accuracy: 0.8505\n",
      "Epoch 135/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.3467 - accuracy: 0.8514 - val_loss: 0.3103 - val_accuracy: 0.8681\n",
      "Epoch 136/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3470 - accuracy: 0.8531 - val_loss: 0.3082 - val_accuracy: 0.8555\n",
      "Epoch 137/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3469 - accuracy: 0.8490 - val_loss: 0.3150 - val_accuracy: 0.8492\n",
      "Epoch 138/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3431 - accuracy: 0.8568 - val_loss: 0.3201 - val_accuracy: 0.8593\n",
      "Epoch 139/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3389 - accuracy: 0.8543 - val_loss: 0.3165 - val_accuracy: 0.8568\n",
      "Epoch 140/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3426 - accuracy: 0.8556 - val_loss: 0.3180 - val_accuracy: 0.8467\n",
      "Epoch 141/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3629 - accuracy: 0.8465 - val_loss: 0.3277 - val_accuracy: 0.8518\n",
      "Epoch 142/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3430 - accuracy: 0.8506 - val_loss: 0.3195 - val_accuracy: 0.8518\n",
      "Epoch 143/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3384 - accuracy: 0.8577 - val_loss: 0.3274 - val_accuracy: 0.8405\n",
      "Epoch 144/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3745 - accuracy: 0.8506 - val_loss: 0.3189 - val_accuracy: 0.8543\n",
      "Epoch 145/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3501 - accuracy: 0.8529 - val_loss: 0.3199 - val_accuracy: 0.8530\n",
      "Epoch 146/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3479 - accuracy: 0.8559 - val_loss: 0.3199 - val_accuracy: 0.8492\n",
      "Epoch 147/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3422 - accuracy: 0.8550 - val_loss: 0.3194 - val_accuracy: 0.8580\n",
      "Epoch 148/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3409 - accuracy: 0.8534 - val_loss: 0.3102 - val_accuracy: 0.8593\n",
      "Epoch 149/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3334 - accuracy: 0.8563 - val_loss: 0.3108 - val_accuracy: 0.8580\n",
      "Epoch 150/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3427 - accuracy: 0.8535 - val_loss: 0.3237 - val_accuracy: 0.8593\n",
      "Epoch 151/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.3378 - accuracy: 0.8570 - val_loss: 0.3137 - val_accuracy: 0.8606\n",
      "Epoch 152/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.3328 - accuracy: 0.8591 - val_loss: 0.3027 - val_accuracy: 0.8668\n",
      "Epoch 153/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3339 - accuracy: 0.8545 - val_loss: 0.3120 - val_accuracy: 0.8568\n",
      "Epoch 154/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3340 - accuracy: 0.8578 - val_loss: 0.3159 - val_accuracy: 0.8580\n",
      "Epoch 155/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.3351 - accuracy: 0.8535 - val_loss: 0.3279 - val_accuracy: 0.8681\n",
      "Epoch 156/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.3394 - accuracy: 0.8482 - val_loss: 0.3295 - val_accuracy: 0.8530\n",
      "Epoch 157/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3335 - accuracy: 0.8588 - val_loss: 0.3413 - val_accuracy: 0.8505\n",
      "Epoch 158/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.3358 - accuracy: 0.8582 - val_loss: 0.3112 - val_accuracy: 0.8668\n",
      "Epoch 159/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3321 - accuracy: 0.8570 - val_loss: 0.3100 - val_accuracy: 0.8631\n",
      "Epoch 160/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3331 - accuracy: 0.8595 - val_loss: 0.3004 - val_accuracy: 0.8568\n",
      "Epoch 161/100000\n",
      "144/144 [==============================] - 0s 3ms/step - loss: 0.3304 - accuracy: 0.8609 - val_loss: 0.3187 - val_accuracy: 0.8555\n",
      "Epoch 162/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3346 - accuracy: 0.8568 - val_loss: 0.3213 - val_accuracy: 0.8530\n",
      "Epoch 163/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3313 - accuracy: 0.8603 - val_loss: 0.3065 - val_accuracy: 0.8580\n",
      "Epoch 164/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3357 - accuracy: 0.8561 - val_loss: 0.3106 - val_accuracy: 0.8555\n",
      "Epoch 165/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3255 - accuracy: 0.8612 - val_loss: 0.2974 - val_accuracy: 0.8606\n",
      "Epoch 166/100000\n",
      "144/144 [==============================] - 1s 4ms/step - loss: 0.3282 - accuracy: 0.8591 - val_loss: 0.3260 - val_accuracy: 0.8568\n",
      "Epoch 167/100000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3315 - accuracy: 0.8573 - val_loss: 0.3026 - val_accuracy: 0.8681\n",
      "Epoch 168/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3322 - accuracy: 0.8594 - val_loss: 0.3453 - val_accuracy: 0.8379\n",
      "Epoch 169/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3281 - accuracy: 0.8617 - val_loss: 0.3056 - val_accuracy: 0.8593\n",
      "Epoch 170/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3282 - accuracy: 0.8609 - val_loss: 0.3212 - val_accuracy: 0.8606\n",
      "Epoch 171/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3348 - accuracy: 0.8529 - val_loss: 0.3206 - val_accuracy: 0.8518\n",
      "Epoch 172/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3262 - accuracy: 0.8610 - val_loss: 0.3130 - val_accuracy: 0.8606\n",
      "Epoch 173/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3260 - accuracy: 0.8620 - val_loss: 0.3145 - val_accuracy: 0.8618\n",
      "Epoch 174/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3264 - accuracy: 0.8577 - val_loss: 0.3092 - val_accuracy: 0.8606\n",
      "Epoch 175/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3296 - accuracy: 0.8598 - val_loss: 0.3062 - val_accuracy: 0.8668\n",
      "Epoch 176/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3283 - accuracy: 0.8620 - val_loss: 0.3030 - val_accuracy: 0.8693\n",
      "Epoch 177/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3305 - accuracy: 0.8567 - val_loss: 0.3088 - val_accuracy: 0.8681\n",
      "Epoch 178/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3269 - accuracy: 0.8624 - val_loss: 0.3094 - val_accuracy: 0.8606\n",
      "Epoch 179/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3308 - accuracy: 0.8575 - val_loss: 0.3149 - val_accuracy: 0.8668\n",
      "Epoch 180/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3320 - accuracy: 0.8566 - val_loss: 0.3039 - val_accuracy: 0.8631\n",
      "Epoch 181/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3246 - accuracy: 0.8610 - val_loss: 0.3240 - val_accuracy: 0.8518\n",
      "Epoch 182/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3289 - accuracy: 0.8645 - val_loss: 0.3070 - val_accuracy: 0.8580\n",
      "Epoch 183/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3265 - accuracy: 0.8641 - val_loss: 0.3221 - val_accuracy: 0.8555\n",
      "Epoch 184/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3293 - accuracy: 0.8584 - val_loss: 0.3190 - val_accuracy: 0.8618\n",
      "Epoch 185/100000\n",
      "144/144 [==============================] - 0s 2ms/step - loss: 0.3353 - accuracy: 0.8585 - val_loss: 0.4035 - val_accuracy: 0.8178\n"
     ]
    }
   ],
   "source": [
    "#fit model\n",
    "history = model.fit(X_train, y_train,\n",
    "                    validation_data = (X_test, y_test),\n",
    "                    callbacks=[es], #early stopping callbacks\n",
    "                    epochs=100000, #max # of epochs for training\n",
    "                    batch_size=50, #hyperparameter\n",
    "                    shuffle=True, #shuffle\n",
    "                    verbose=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Evaluate Model\n",
    "## Question 9"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 323,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#loss for training and validation sets from training history\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "#X range\n",
    "epochs = range(1, len(acc) + 1)\n",
    "\n",
    "#plot training loss, 'r' for red line\n",
    "plt.plot(epochs, loss, 'r', label='Training loss')\n",
    "\n",
    "#plot validation loss, 'b'for blue line\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 324,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#accuracy for training and validation sets from training history\n",
    "acc = history.history['accuracy']\n",
    "val_acc = history.history['val_accuracy']\n",
    "\n",
    "#X range\n",
    "epochs = range(1, len(loss_values) + 1)\n",
    "\n",
    "#plot training accuracy, 'b' for blue line\n",
    "plt.plot(epochs, acc, 'b', label='Training accuracy')\n",
    "\n",
    "#plot validation accuracy, 'r'for red line\n",
    "plt.plot(epochs, val_acc, 'r', label='Validation accuracy')\n",
    "\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 325,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8706030249595642"
      ]
     },
     "execution_count": 325,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#highest validation accuracy\n",
    "np.max(val_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 310,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics import classification_report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 311,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "249/249 [==============================] - 0s 1ms/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[9.9841267e-01],\n",
       "       [9.9637330e-01],\n",
       "       [9.9052656e-01],\n",
       "       ...,\n",
       "       [9.1531223e-01],\n",
       "       [9.1045594e-04],\n",
       "       [6.6519076e-01]], dtype=float32)"
      ]
     },
     "execution_count": 311,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.predict(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 312,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "249/249 [==============================] - 0s 1ms/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[1.],\n",
       "       [1.],\n",
       "       [1.],\n",
       "       ...,\n",
       "       [1.],\n",
       "       [0.],\n",
       "       [1.]], dtype=float32)"
      ]
     },
     "execution_count": 312,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.round(model.predict(X),0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 313,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0       1\n",
       "1       1\n",
       "2       1\n",
       "3       1\n",
       "4       1\n",
       "       ..\n",
       "8028    1\n",
       "8029    0\n",
       "8030    0\n",
       "8031    0\n",
       "8032    0\n",
       "Name: Poverty_recode, Length: 7956, dtype: int64"
      ]
     },
     "execution_count": 313,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 314,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "249/249 [==============================] - 0s 1ms/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[3941,  379],\n",
       "       [ 623, 3013]])"
      ]
     },
     "execution_count": 314,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds = np.round(model.predict(X),0)\n",
    "\n",
    "#confusion matrix\n",
    "confusion_matrix(y, preds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 315,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.86      0.91      0.89      4320\n",
      "           1       0.89      0.83      0.86      3636\n",
      "\n",
      "    accuracy                           0.87      7956\n",
      "   macro avg       0.88      0.87      0.87      7956\n",
      "weighted avg       0.87      0.87      0.87      7956\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#classification report\n",
    "print(classification_report(y, preds))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The model started with a low accuracy. The model started to improve at around the 20th epoch. The model did learn from the data but not gently.\n",
    "\n",
    "The validation accuracy peaked at 0.87. It plateaued at this point. \n",
    "\n",
    "The loss values fluctuate at the start and then stabalize as the training progresses. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Baseline Model\n",
    "## Question 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 326,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "#initialize LR model\n",
    "LR = LogisticRegression(max_iter=100000000)\n",
    "\n",
    "#fit to training data\n",
    "clf = LR.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 327,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[3896  424]\n",
      " [ 654 2982]]\n"
     ]
    }
   ],
   "source": [
    "preds = clf.predict(X)\n",
    "\n",
    "#confusion matrix\n",
    "print(confusion_matrix(y, preds))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 328,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.86      0.90      0.88      4320\n",
      "           1       0.88      0.82      0.85      3636\n",
      "\n",
      "    accuracy                           0.86      7956\n",
      "   macro avg       0.87      0.86      0.86      7956\n",
      "weighted avg       0.87      0.86      0.86      7956\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#classification report\n",
    "print(classification_report(y, preds))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The NN model did better than the baseline LR model. The NN model had slightly higher numbers in all error metrics. The NN model had more true positives and true negatives and less false negatives and false poistives. "
   ]
  }
 ],
 "metadata": {
  "colab": {
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}