analysis notebooks/LivingIncome_Model.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Living Income Analysis: Machine Learning Model with max 10 features"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Summary:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this notebook I develop a Restricted Machine Learning model that can predict whether a cocoa farming household in Ghana and Côte d'Ivoire reach a Living Income. The Machine Learning model is restricted to 10 features, to make its real-world application easy and cost-effective.\n",
    "\n",
    "I use as starting point the Machine Learning model prepared at [Living Income Analysis: Machine Learning notebook](LivingIncome_MachineLearning.ipynb), which in turn used\n",
    "as starting point the dataset prepared at [Living Income Analysis notebook](LivingIncomeAnalysis.ipynb)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this notebook, specifically I will:\n",
    "\n",
    "* Load the dataset\n",
    "* Extract the top 10 features\n",
    "* Re-fit the best classifier with the restricted dataset, using Grid Search with a stratified sross-validation\n",
    "* Search broadly using TPOT, with the same stratified cross-validation set \n",
    "* Compare performance and choose the best classifier with the restricted data\n",
    "* Calibrate the probabilities\n",
    "* Export the final model such that it can be used in a web-app and by others"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 0 Preamble"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 0.1 Load libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\marce\\anaconda3\\envs\\livingincomemodel\\lib\\site-packages\\tpot\\builtins\\__init__.py:36: UserWarning: Warning: optional dependency `torch` is not available. - skipping import of NN models.\n",
      "  warnings.warn(\"Warning: optional dependency `torch` is not available. - skipping import of NN models.\")\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import pickle\n",
    "from time import time\n",
    "\n",
    "from sklearn.ensemble import AdaBoostClassifier\n",
    "\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.metrics import fbeta_score\n",
    "from sklearn.metrics import make_scorer\n",
    "from sklearn.metrics import precision_recall_fscore_support\n",
    "\n",
    "from sklearn.metrics import plot_confusion_matrix\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics import ConfusionMatrixDisplay\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import StratifiedShuffleSplit\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "from tpot import TPOTClassifier\n",
    "\n",
    "from sklearn.calibration import CalibratedClassifierCV\n",
    "\n",
    "import visuals as vs # this is a slightly modifed code which provided via de Udacity Nanodegree Intro to Machine Learning"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 0.1 Load Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load data as in the previous notebook [Living Income Analysis: Machine Learning notebook](LivingIncome_MachineLearning.ipynb)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_pickle('../data/data_ready_for_ML.pkl')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Apply the same strategy to fill missing values, so the data is identical:\n",
    "\n",
    "* For categorical variables, use the mode\n",
    "* For numerical variable, use the median\n",
    "\n",
    "The strategy above is also quite robust to outliers, as the mode and median are typically quite stable and not sensitive to extreme values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# iterate through all features\n",
    "for this_column in list(data.columns):\n",
    "\n",
    "    # check if categorical\n",
    "    if (str(data[this_column].dtype) == 'category') | (str(data[this_column].dtype) == 'object') :\n",
    "        fill_value = data[this_column].mode()\n",
    "\n",
    "    else:\n",
    "        fill_value = data[this_column].median()\n",
    "     \n",
    "    # fill missing values\n",
    "    data[this_column] = data[this_column].fillna(value = fill_value)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Split into features and target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "target = data['Living Income Achieved']\n",
    "data.drop(['Living Income Achieved'], axis = 1, inplace = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Expand dummies, to maintain the original shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.get_dummies(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Check shape, which should be `(1377, 838)`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1377, 838)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 0.2 Load Best ML Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load the best model from the previous notebook [Living Income Analysis: Machine Learning notebook](LivingIncome_MachineLearning.ipynb)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "best_model = pd.read_pickle('../models/best_overall.pkl')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Get the 10 best features:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>Feature weight</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Estimated income percentage from sale of cocoa</th>\n",
       "      <td>0.092667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Cocoa production (kg/ha)</th>\n",
       "      <td>0.090667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Cocoa production (kg)</th>\n",
       "      <td>0.059333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Productive land under cocoa (ha)</th>\n",
       "      <td>0.042000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Cocoa land owned (ha)</th>\n",
       "      <td>0.029333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Hectares with trees between 5 and 25 years old</th>\n",
       "      <td>0.022000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Estimated income percentage from own small business or trading</th>\n",
       "      <td>0.021333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Head: age</th>\n",
       "      <td>0.020667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Estimated income percentage from sale of other crops</th>\n",
       "      <td>0.020000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Land used to cultivate all crops (ha)</th>\n",
       "      <td>0.014667</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                    Feature weight\n",
       "Estimated income percentage from sale of cocoa            0.092667\n",
       "Cocoa production (kg/ha)                                  0.090667\n",
       "Cocoa production (kg)                                     0.059333\n",
       "Productive land under cocoa (ha)                          0.042000\n",
       "Cocoa land owned (ha)                                     0.029333\n",
       "Hectares with trees between 5 and 25 years old            0.022000\n",
       "Estimated income percentage from own small busi...        0.021333\n",
       "Head: age                                                 0.020667\n",
       "Estimated income percentage from sale of other ...        0.020000\n",
       "Land used to cultivate all crops (ha)                     0.014667"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_10_features = pd.DataFrame({'Feature weight': best_model.feature_importances_}, index=data.columns).sort_values(by='Feature weight', ascending=False).head(n=10)\n",
    "top_10_features"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Plot the top 10 features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x1500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "vs.feature_plot(best_model.feature_importances_, data, target, n_features= 10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Store previous performance metrics for comparison"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "FULL_adaOptimStrat_accuracy = 0.9652\n",
    "FULL_adaOptimStrat_fscore = 0.8667"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ADAboost, Full Data Optimized Stratified Predictor: [Accuracy score: 0.9652, F-score: 0.8667]\n"
     ]
    }
   ],
   "source": [
    "print(\"ADAboost, Full Data Optimized Stratified Predictor: [Accuracy score: {:.4f}, F-score: {:.4f}]\".format(FULL_adaOptimStrat_accuracy, FULL_adaOptimStrat_fscore))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1. Get Data ready"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Extract top 10 features:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_top10 = data[top_10_features.index]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Confirm shape:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1377, 10)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_top10.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Check basic descriptives"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "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>Estimated income percentage from sale of cocoa</th>\n",
       "      <th>Cocoa production (kg/ha)</th>\n",
       "      <th>Cocoa production (kg)</th>\n",
       "      <th>Productive land under cocoa (ha)</th>\n",
       "      <th>Cocoa land owned (ha)</th>\n",
       "      <th>Hectares with trees between 5 and 25 years old</th>\n",
       "      <th>Estimated income percentage from own small business or trading</th>\n",
       "      <th>Head: age</th>\n",
       "      <th>Estimated income percentage from sale of other crops</th>\n",
       "      <th>Land used to cultivate all crops (ha)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1377.000000</td>\n",
       "      <td>1377.000000</td>\n",
       "      <td>1377.000000</td>\n",
       "      <td>1377.000000</td>\n",
       "      <td>1377.000000</td>\n",
       "      <td>1377.000000</td>\n",
       "      <td>1377.000000</td>\n",
       "      <td>1377.000000</td>\n",
       "      <td>1377.000000</td>\n",
       "      <td>1377.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.636100</td>\n",
       "      <td>401.369507</td>\n",
       "      <td>1103.353516</td>\n",
       "      <td>3.062009</td>\n",
       "      <td>3.088592</td>\n",
       "      <td>2.452407</td>\n",
       "      <td>0.077807</td>\n",
       "      <td>50.591140</td>\n",
       "      <td>0.205309</td>\n",
       "      <td>5.279386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.208974</td>\n",
       "      <td>239.152908</td>\n",
       "      <td>947.676453</td>\n",
       "      <td>2.325433</td>\n",
       "      <td>2.597953</td>\n",
       "      <td>2.076500</td>\n",
       "      <td>0.132000</td>\n",
       "      <td>13.217897</td>\n",
       "      <td>0.170920</td>\n",
       "      <td>3.430948</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.050000</td>\n",
       "      <td>20.000000</td>\n",
       "      <td>20.000000</td>\n",
       "      <td>0.202343</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>16.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.352800</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.500000</td>\n",
       "      <td>221.036102</td>\n",
       "      <td>447.999969</td>\n",
       "      <td>1.500000</td>\n",
       "      <td>1.500000</td>\n",
       "      <td>1.011716</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>41.000000</td>\n",
       "      <td>0.100000</td>\n",
       "      <td>2.832804</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.650000</td>\n",
       "      <td>355.831177</td>\n",
       "      <td>768.000000</td>\n",
       "      <td>2.428118</td>\n",
       "      <td>2.450000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>50.000000</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>4.249603</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.800000</td>\n",
       "      <td>525.000000</td>\n",
       "      <td>1478.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>4.046863</td>\n",
       "      <td>3.237490</td>\n",
       "      <td>0.100000</td>\n",
       "      <td>60.000000</td>\n",
       "      <td>0.300000</td>\n",
       "      <td>6.717792</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1423.324829</td>\n",
       "      <td>5600.000000</td>\n",
       "      <td>15.000000</td>\n",
       "      <td>15.000000</td>\n",
       "      <td>13.750000</td>\n",
       "      <td>0.900000</td>\n",
       "      <td>95.000000</td>\n",
       "      <td>0.900000</td>\n",
       "      <td>21.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Estimated income percentage from sale of cocoa  \\\n",
       "count                                     1377.000000   \n",
       "mean                                         0.636100   \n",
       "std                                          0.208974   \n",
       "min                                          0.050000   \n",
       "25%                                          0.500000   \n",
       "50%                                          0.650000   \n",
       "75%                                          0.800000   \n",
       "max                                          1.000000   \n",
       "\n",
       "       Cocoa production (kg/ha)  Cocoa production (kg)  \\\n",
       "count               1377.000000            1377.000000   \n",
       "mean                 401.369507            1103.353516   \n",
       "std                  239.152908             947.676453   \n",
       "min                   20.000000              20.000000   \n",
       "25%                  221.036102             447.999969   \n",
       "50%                  355.831177             768.000000   \n",
       "75%                  525.000000            1478.000000   \n",
       "max                 1423.324829            5600.000000   \n",
       "\n",
       "       Productive land under cocoa (ha)  Cocoa land owned (ha)  \\\n",
       "count                       1377.000000            1377.000000   \n",
       "mean                           3.062009               3.088592   \n",
       "std                            2.325433               2.597953   \n",
       "min                            0.202343               0.000000   \n",
       "25%                            1.500000               1.500000   \n",
       "50%                            2.428118               2.450000   \n",
       "75%                            4.000000               4.046863   \n",
       "max                           15.000000              15.000000   \n",
       "\n",
       "       Hectares with trees between 5 and 25 years old  \\\n",
       "count                                     1377.000000   \n",
       "mean                                         2.452407   \n",
       "std                                          2.076500   \n",
       "min                                          0.000000   \n",
       "25%                                          1.011716   \n",
       "50%                                          2.000000   \n",
       "75%                                          3.237490   \n",
       "max                                         13.750000   \n",
       "\n",
       "       Estimated income percentage from own small business or trading  \\\n",
       "count                                        1377.000000                \n",
       "mean                                            0.077807                \n",
       "std                                             0.132000                \n",
       "min                                             0.000000                \n",
       "25%                                             0.000000                \n",
       "50%                                             0.000000                \n",
       "75%                                             0.100000                \n",
       "max                                             0.900000                \n",
       "\n",
       "         Head: age  Estimated income percentage from sale of other crops  \\\n",
       "count  1377.000000                                        1377.000000      \n",
       "mean     50.591140                                           0.205309      \n",
       "std      13.217897                                           0.170920      \n",
       "min      16.000000                                           0.000000      \n",
       "25%      41.000000                                           0.100000      \n",
       "50%      50.000000                                           0.200000      \n",
       "75%      60.000000                                           0.300000      \n",
       "max      95.000000                                           0.900000      \n",
       "\n",
       "       Land used to cultivate all crops (ha)  \n",
       "count                            1377.000000  \n",
       "mean                                5.279386  \n",
       "std                                 3.430948  \n",
       "min                                 0.352800  \n",
       "25%                                 2.832804  \n",
       "50%                                 4.249603  \n",
       "75%                                 6.717792  \n",
       "max                                21.000000  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_top10.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Export the data used for the restricted model:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_for_restricted_ML = pd.concat([target, data_top10], axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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>Living Income Achieved</th>\n",
       "      <th>Estimated income percentage from sale of cocoa</th>\n",
       "      <th>Cocoa production (kg/ha)</th>\n",
       "      <th>Cocoa production (kg)</th>\n",
       "      <th>Productive land under cocoa (ha)</th>\n",
       "      <th>Cocoa land owned (ha)</th>\n",
       "      <th>Hectares with trees between 5 and 25 years old</th>\n",
       "      <th>Estimated income percentage from own small business or trading</th>\n",
       "      <th>Head: age</th>\n",
       "      <th>Estimated income percentage from sale of other crops</th>\n",
       "      <th>Land used to cultivate all crops (ha)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>False</td>\n",
       "      <td>0.89</td>\n",
       "      <td>177.915604</td>\n",
       "      <td>576.0</td>\n",
       "      <td>3.237490</td>\n",
       "      <td>4.046863</td>\n",
       "      <td>3.237490</td>\n",
       "      <td>0.00</td>\n",
       "      <td>49.0</td>\n",
       "      <td>0.11</td>\n",
       "      <td>4.856235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>False</td>\n",
       "      <td>0.60</td>\n",
       "      <td>722.958618</td>\n",
       "      <td>1024.0</td>\n",
       "      <td>1.416402</td>\n",
       "      <td>1.618745</td>\n",
       "      <td>1.416402</td>\n",
       "      <td>0.00</td>\n",
       "      <td>63.0</td>\n",
       "      <td>0.25</td>\n",
       "      <td>3.035147</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>False</td>\n",
       "      <td>0.50</td>\n",
       "      <td>210.862946</td>\n",
       "      <td>128.0</td>\n",
       "      <td>0.607029</td>\n",
       "      <td>0.607029</td>\n",
       "      <td>0.607029</td>\n",
       "      <td>0.35</td>\n",
       "      <td>53.0</td>\n",
       "      <td>0.15</td>\n",
       "      <td>0.809373</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>False</td>\n",
       "      <td>0.65</td>\n",
       "      <td>316.294403</td>\n",
       "      <td>512.0</td>\n",
       "      <td>1.618745</td>\n",
       "      <td>1.618745</td>\n",
       "      <td>1.618745</td>\n",
       "      <td>0.00</td>\n",
       "      <td>65.0</td>\n",
       "      <td>0.30</td>\n",
       "      <td>2.428118</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>False</td>\n",
       "      <td>0.80</td>\n",
       "      <td>197.683990</td>\n",
       "      <td>320.0</td>\n",
       "      <td>1.618745</td>\n",
       "      <td>2.428118</td>\n",
       "      <td>1.618745</td>\n",
       "      <td>0.00</td>\n",
       "      <td>60.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>2.428118</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Living Income Achieved  Estimated income percentage from sale of cocoa  \\\n",
       "0                   False                                            0.89   \n",
       "1                   False                                            0.60   \n",
       "2                   False                                            0.50   \n",
       "3                   False                                            0.65   \n",
       "5                   False                                            0.80   \n",
       "\n",
       "   Cocoa production (kg/ha)  Cocoa production (kg)  \\\n",
       "0                177.915604                  576.0   \n",
       "1                722.958618                 1024.0   \n",
       "2                210.862946                  128.0   \n",
       "3                316.294403                  512.0   \n",
       "5                197.683990                  320.0   \n",
       "\n",
       "   Productive land under cocoa (ha)  Cocoa land owned (ha)  \\\n",
       "0                          3.237490               4.046863   \n",
       "1                          1.416402               1.618745   \n",
       "2                          0.607029               0.607029   \n",
       "3                          1.618745               1.618745   \n",
       "5                          1.618745               2.428118   \n",
       "\n",
       "   Hectares with trees between 5 and 25 years old  \\\n",
       "0                                        3.237490   \n",
       "1                                        1.416402   \n",
       "2                                        0.607029   \n",
       "3                                        1.618745   \n",
       "5                                        1.618745   \n",
       "\n",
       "   Estimated income percentage from own small business or trading  Head: age  \\\n",
       "0                                               0.00                    49.0   \n",
       "1                                               0.00                    63.0   \n",
       "2                                               0.35                    53.0   \n",
       "3                                               0.00                    65.0   \n",
       "5                                               0.00                    60.0   \n",
       "\n",
       "   Estimated income percentage from sale of other crops  \\\n",
       "0                                               0.11      \n",
       "1                                               0.25      \n",
       "2                                               0.15      \n",
       "3                                               0.30      \n",
       "5                                               0.00      \n",
       "\n",
       "   Land used to cultivate all crops (ha)  \n",
       "0                               4.856235  \n",
       "1                               3.035147  \n",
       "2                               0.809373  \n",
       "3                               2.428118  \n",
       "5                               2.428118  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_for_restricted_ML.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_for_restricted_ML.to_pickle('../data/data_for_restricted_ML.pkl')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2. Grid Search best parameters, using stratified shuffle split"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Split data in training and test set:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(data_top10, target,\n",
    "                                                    train_size=0.75, test_size=0.25,\n",
    "                                                   random_state = 32)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.1 Re-fit AdaBoostClassifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Prepare elements for the grid search. using the same parameters as in the previous notebook"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# base classifier\n",
    "clf = AdaBoostClassifier(random_state=42)\n",
    "\n",
    "# Create the parameters list to tune\n",
    "parameters = {'n_estimators':[10, 50, 500, 1200, 1500, 2000],\n",
    "              'learning_rate':[1.0,1.5,2.0]}\n",
    "\n",
    "# Make an fbeta_score scoring object using make_scorer()\n",
    "scorer = make_scorer(fbeta_score, beta= 0.5)\n",
    "\n",
    "# Make sure it is stratified\n",
    "cv_strat = StratifiedShuffleSplit(n_splits=5,\n",
    "                                  test_size=0.2,\n",
    "                                  random_state=42)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Run the grid search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 18 candidates, totalling 90 fits\n",
      "Best model:\n",
      "AdaBoostClassifier(n_estimators=500, random_state=42)\n",
      "------\n",
      "\n",
      "Restricted Optimized Model with stratification\n",
      "------\n",
      "Final accuracy score on the testing data: 0.9536\n",
      "Final F-score on the testing data: 0.8458\n"
     ]
    }
   ],
   "source": [
    "## Create a Grid Search instance, using stratified CV\n",
    "grid_obj_strat = GridSearchCV(clf, parameters, scoring = scorer, verbose=2,\n",
    "                             cv = cv_strat, n_jobs = -1)\n",
    "\n",
    "# Fit the grid search object to the training data and find the optimal parameters using fit()\n",
    "grid_fit_strat = grid_obj_strat.fit(X_train, y_train)\n",
    "\n",
    "# Get the estimator\n",
    "best_clf_strat = grid_fit_strat.best_estimator_\n",
    "\n",
    "# print best model:\n",
    "print(\"Best model:\")\n",
    "print(best_clf_strat)\n",
    "print(\"------\")\n",
    "\n",
    "# Make predictions \n",
    "best_predictions_strat = best_clf_strat.predict(X_test)\n",
    "\n",
    "# metrics\n",
    "adaOptimStrat_accuracy = accuracy_score(y_test, best_predictions_strat)\n",
    "adaOptimStrat_fscore = fbeta_score(y_test, best_predictions_strat, beta = 0.5)\n",
    "\n",
    "print(\"\\nRestricted Optimized Model with stratification\\n------\")\n",
    "print(\"Final accuracy score on the testing data: {:.4f}\".format(adaOptimStrat_accuracy))\n",
    "print(\"Final F-score on the testing data: {:.4f}\".format(adaOptimStrat_fscore))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Because the solution is on the limits (`learning rate = 1.0`), let's expand grid parameters:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "parameters = {'n_estimators':[10, 50, 500, 1200, 1300, 1500, 1600, 2000],\n",
    "              'learning_rate':[0.05, 0.1, 0.2, 0.3, 0.5,1.0,1.5,2.0]}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Re-run Grid Search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 64 candidates, totalling 320 fits\n",
      "Best model:\n",
      "AdaBoostClassifier(learning_rate=0.1, n_estimators=1500, random_state=42)\n",
      "------\n",
      "\n",
      "Restricted Optimized Model with stratification\n",
      "------\n",
      "Final accuracy score on the testing data: 0.9623\n",
      "Final F-score on the testing data: 0.8780\n"
     ]
    }
   ],
   "source": [
    "## Create a Grid Search instance, using stratified CV\n",
    "grid_obj_strat = GridSearchCV(clf, parameters, scoring = scorer, verbose=2,\n",
    "                             cv = cv_strat, n_jobs = -1)\n",
    "\n",
    "# Fit the grid search object to the training data and find the optimal parameters using fit()\n",
    "grid_fit_strat = grid_obj_strat.fit(X_train, y_train)\n",
    "\n",
    "# Get the estimator\n",
    "best_clf_strat = grid_fit_strat.best_estimator_\n",
    "\n",
    "# print best model:\n",
    "print(\"Best model:\")\n",
    "print(best_clf_strat)\n",
    "print(\"------\")\n",
    "\n",
    "# Make predictions \n",
    "best_predictions_strat = best_clf_strat.predict(X_test)\n",
    "\n",
    "# metrics\n",
    "adaOptimStrat_accuracy = accuracy_score(y_test, best_predictions_strat)\n",
    "adaOptimStrat_fscore = fbeta_score(y_test, best_predictions_strat, beta = 0.5)\n",
    "\n",
    "print(\"\\nRestricted Optimized Model with stratification\\n------\")\n",
    "print(\"Final accuracy score on the testing data: {:.4f}\".format(adaOptimStrat_accuracy))\n",
    "print(\"Final F-score on the testing data: {:.4f}\".format(adaOptimStrat_fscore))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compare full with the restricted model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ADAboost, Full Data Optimized Stratified Predictor: [Accuracy score: 0.9652, F-score: 0.8667]\n",
      "ADAboost, Restricted Data Optimized Stratified Predictor: [Accuracy score: 0.9623, F-score: 0.8780]\n"
     ]
    }
   ],
   "source": [
    "print(\"ADAboost, Full Data Optimized Stratified Predictor: [Accuracy score: {:.4f}, F-score: {:.4f}]\".format(FULL_adaOptimStrat_accuracy, FULL_adaOptimStrat_fscore))\n",
    "print(\"ADAboost, Restricted Data Optimized Stratified Predictor: [Accuracy score: {:.4f}, F-score: {:.4f}]\".format(adaOptimStrat_accuracy, adaOptimStrat_fscore))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Check broader performance metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([0.9704918, 0.9      ]),\n",
       " array([0.98666667, 0.8       ]),\n",
       " array([0.97368421, 0.87804878]),\n",
       " array([300,  45], dtype=int64))"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "precision_recall_fscore_support(y_test, best_predictions_strat, beta = 0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[296,   4],\n",
       "       [  9,  36]], dtype=int64)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "confusion_matrix(y_test, best_predictions_strat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\marce\\anaconda3\\envs\\livingincomemodel\\lib\\site-packages\\sklearn\\utils\\deprecation.py:87: FutureWarning: Function plot_confusion_matrix is deprecated; Function `plot_confusion_matrix` is deprecated in 1.0 and will be removed in 1.2. Use one of the class methods: ConfusionMatrixDisplay.from_predictions or ConfusionMatrixDisplay.from_estimator.\n",
      "  warnings.warn(msg, category=FutureWarning)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_confusion_matrix(best_clf_strat, \n",
    "                      X_test, y_test,\n",
    "                      cmap=plt.cm.Blues,\n",
    "                      normalize = 'true');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's save the best model so far:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "pkl_filename = '../models/restricted_ada_best_strat.pkl'\n",
    "pickle.dump(best_clf_strat, open(pkl_filename, 'wb'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.1 Search broadly using TPOT"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "pipeline_optimizer = TPOTClassifier()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "pipeline_optimizer = TPOTClassifier(verbosity=2,\n",
    "                                     n_jobs = -1,\n",
    "                                    random_state = 32,\n",
    "                                   periodic_checkpoint_folder = \"../models/intermediate_results\",\n",
    "                                    cv = cv_strat,\n",
    "                                    scoring = scorer\n",
    "                                   )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Optimization Progress', max=10100.0, style=ProgressStyle(…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Generation 1 - Current best internal CV score: 0.8950299281181634\n",
      "\n",
      "Generation 2 - Current best internal CV score: 0.8950299281181634\n",
      "\n",
      "Generation 3 - Current best internal CV score: 0.8950299281181634\n",
      "\n",
      "Generation 4 - Current best internal CV score: 0.8950299281181634\n",
      "\n",
      "Generation 5 - Current best internal CV score: 0.8969206844206845\n",
      "\n",
      "Generation 6 - Current best internal CV score: 0.8969206844206845\n",
      "\n",
      "Generation 7 - Current best internal CV score: 0.8973901401693443\n",
      "\n",
      "Generation 8 - Current best internal CV score: 0.9036664724164722\n",
      "\n",
      "Generation 9 - Current best internal CV score: 0.9036664724164722\n",
      "\n",
      "Generation 10 - Current best internal CV score: 0.9036664724164722\n",
      "\n",
      "Generation 11 - Current best internal CV score: 0.9036664724164722\n",
      "\n",
      "Generation 12 - Current best internal CV score: 0.9116475346314055\n",
      "\n",
      "Generation 13 - Current best internal CV score: 0.9116475346314055\n",
      "\n",
      "Generation 14 - Current best internal CV score: 0.9116475346314055\n",
      "\n",
      "Generation 15 - Current best internal CV score: 0.9161531279178338\n",
      "\n",
      "Generation 16 - Current best internal CV score: 0.9223026598026598\n",
      "\n",
      "Generation 17 - Current best internal CV score: 0.9223026598026598\n",
      "\n",
      "Generation 18 - Current best internal CV score: 0.9223026598026598\n",
      "\n",
      "Generation 19 - Current best internal CV score: 0.9223026598026598\n",
      "\n",
      "Generation 20 - Current best internal CV score: 0.9223026598026598\n",
      "\n",
      "Generation 21 - Current best internal CV score: 0.9223026598026598\n",
      "\n",
      "Generation 22 - Current best internal CV score: 0.9223026598026598\n",
      "\n",
      "Generation 23 - Current best internal CV score: 0.9223026598026598\n",
      "\n",
      "Generation 24 - Current best internal CV score: 0.9255706336588689\n",
      "\n",
      "Generation 25 - Current best internal CV score: 0.9255706336588689\n",
      "\n",
      "Generation 26 - Current best internal CV score: 0.9255706336588689\n",
      "\n",
      "Generation 27 - Current best internal CV score: 0.9255706336588689\n",
      "\n",
      "Generation 28 - Current best internal CV score: 0.9255706336588689\n",
      "\n",
      "Generation 29 - Current best internal CV score: 0.9255706336588689\n",
      "\n",
      "Generation 30 - Current best internal CV score: 0.9255706336588689\n",
      "\n",
      "Generation 31 - Current best internal CV score: 0.9255706336588689\n",
      "\n",
      "Generation 32 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 33 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 34 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 35 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 36 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 37 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 38 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 39 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 40 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 41 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 42 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 43 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 44 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 45 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 46 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 47 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 48 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 49 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 50 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 51 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 52 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 53 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 54 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 55 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 56 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 57 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 58 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 59 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 60 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 61 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 62 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 63 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 64 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 65 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 66 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 67 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 68 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 69 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 70 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 71 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 72 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 73 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 74 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 75 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 76 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 77 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 78 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 79 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 80 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 81 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 82 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 83 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 84 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 85 - Current best internal CV score: 0.92746138996139\n",
      "\n",
      "Generation 86 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 87 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 88 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 89 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 90 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 91 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 92 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 93 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 94 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 95 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 96 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 97 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 98 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 99 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Generation 100 - Current best internal CV score: 0.932505472401108\n",
      "\n",
      "Best pipeline: GradientBoostingClassifier(Normalizer(RFE(BernoulliNB(input_matrix, alpha=100.0, fit_prior=False), criterion=entropy, max_features=0.8, n_estimators=100, step=0.9000000000000001), norm=l1), learning_rate=0.01, max_depth=2, max_features=0.9500000000000001, min_samples_leaf=7, min_samples_split=19, n_estimators=100, subsample=0.3)\n",
      "0.9116022099447513\n"
     ]
    }
   ],
   "source": [
    "pipeline_optimizer.fit(X_train, y_train)\n",
    "print(pipeline_optimizer.score(X_test, y_test))\n",
    "pipeline_optimizer.export('../models/restricted_tpot_exported_pipeline.py')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Check score:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9116022099447513\n"
     ]
    }
   ],
   "source": [
    "print(pipeline_optimizer.score(X_test, y_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compute performance metrics:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "tpot_pred = pipeline_optimizer.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "# metrics\n",
    "tpot_accuracy = accuracy_score(y_test, tpot_pred)\n",
    "tpot_fscore = fbeta_score(y_test, tpot_pred, beta = 0.5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compare ML models:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ADAboost, Full Data Optimized Stratified Predictor: [Accuracy score: 0.9652, F-score: 0.8667]\n",
      "ADAboost, Restricted Data Optimized Stratified Predictor: [Accuracy score: 0.9623, F-score: 0.8780]\n",
      "TPOT Predictor, Restricted Model: [Accuracy score: 0.9623, F-score: 0.9116]\n"
     ]
    }
   ],
   "source": [
    "print(\"ADAboost, Full Data Optimized Stratified Predictor: [Accuracy score: {:.4f}, F-score: {:.4f}]\".format(FULL_adaOptimStrat_accuracy, FULL_adaOptimStrat_fscore))\n",
    "print(\"ADAboost, Restricted Data Optimized Stratified Predictor: [Accuracy score: {:.4f}, F-score: {:.4f}]\".format(adaOptimStrat_accuracy, adaOptimStrat_fscore))\n",
    "print(\"TPOT Predictor, Restricted Model: [Accuracy score: {:.4f}, F-score: {:.4f}]\".format(tpot_accuracy, tpot_fscore))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([0.96141479, 0.97058824]),\n",
       " array([0.99666667, 0.73333333]),\n",
       " array([0.96826425, 0.91160221]),\n",
       " array([300,  45], dtype=int64))"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "precision_recall_fscore_support(y_test, tpot_pred, beta = 0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[299,   1],\n",
       "       [ 12,  33]], dtype=int64)"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "confusion_matrix(y_test, tpot_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm = confusion_matrix(y_test, tpot_pred, normalize = 'true')\n",
    "\n",
    "cm_display = ConfusionMatrixDisplay(cm).plot(cmap=plt.cm.Blues)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Save best TPOT model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "TPOT_exported_pipeline = pipeline_optimizer.fitted_pipeline_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "tpot_results = TPOT_exported_pipeline.predict(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Double check:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([0.96141479, 0.97058824]),\n",
       " array([0.99666667, 0.73333333]),\n",
       " array([0.96826425, 0.91160221]),\n",
       " array([300,  45], dtype=int64))"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "precision_recall_fscore_support(y_test, tpot_results, beta = 0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "pkl_filename = '../models/restricted_tpot_best_fitted.pkl'\n",
    "pickle.dump(TPOT_exported_pipeline, open(pkl_filename, 'wb'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 7 Model selection"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's compare the best models:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "|           | ADAboost        |         | TPOT            |         |\n",
    "|-----------|-----------------|---------|-----------------|---------|\n",
    "| accuracy  | 96.23%           |         | 96.23%           |         |\n",
    "|           | _Did not achieve_ | _Achieved_ | _Did not Achieve_ | _Achieved_ |\n",
    "| precision | 97%             | 90%     | 96%             | 97%     |\n",
    "| recall    | 98.7%             | 80%     | 99.6%             | 73%     |\n",
    "| f-score   | 97.3%             | 87.8%     | 96.8%           | 91.2%   |"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "TPOT has better _precision_ when it comes to the farmers who _achieved_ the living income benchmark (i.e., from the 34 predictions from the TPOT model, 33 where correct; the ADAboost made 40 predictions, 36 correct).\n",
    "\n",
    "On the other hand, ADAboost has better _recall_ (i.e., from the 45 farmers that _achieved_ the living income benchmark, 36 were found by the ADAboost, while only 33 where found by the TPOT model)\n",
    "\n",
    "Because the ADAboost is more balanced in precision and recall, I prefer this model over the TPOT best pipeline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "best_model =  best_clf_strat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "pkl_filename = '../models/restricted_best_overall.pkl'\n",
    "pickle.dump(best_model, open(pkl_filename, 'wb'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 8 Probabilities"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Besides predicting if a farmer reaches a living income, I'm interested in predicting the probability that a farmer reaches a living income. I do this by calibrating the original model `proba`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For reference see [probability calibration](https://scikit-learn.org/stable/modules/calibration.html#calibration) and this cross-validated [post](https://stats.stackexchange.com/questions/110981/why-is-adaboost-predicting-probabilities-with-so-little-standard-deviation)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 8.1 AdaBoostClassifier "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is the average chance that an observation is classified as reaching a living income (`true`)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.11594202898550725"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(best_clf_strat.predict(X_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "However, this is the average probability that an observation reaches the living income (`proba`) before calibration"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.47577319151783504"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(best_clf_strat.predict_proba(X_test)[:,1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Which are clearly not in line."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To adjust this, we calibrate the classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "calibrated_best_clf_strat = CalibratedClassifierCV(best_clf_strat, cv=cv_strat, method='isotonic')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-1 {color: black;background-color: white;}#sk-container-id-1 pre{padding: 0;}#sk-container-id-1 div.sk-toggleable {background-color: white;}#sk-container-id-1 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-1 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-1 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-1 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-1 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-1 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-1 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-1 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-1 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-1 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-1 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-1 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-1 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-1 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-1 div.sk-item {position: relative;z-index: 1;}#sk-container-id-1 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-1 div.sk-item::before, #sk-container-id-1 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-1 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-1 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-1 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-1 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-1 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-1 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-1 div.sk-label-container {text-align: center;}#sk-container-id-1 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-1 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-1\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>CalibratedClassifierCV(base_estimator=AdaBoostClassifier(learning_rate=0.1,\n",
       "                                                         n_estimators=1500,\n",
       "                                                         random_state=42),\n",
       "                       cv=StratifiedShuffleSplit(n_splits=5, random_state=42, test_size=0.2,\n",
       "            train_size=None),\n",
       "                       method=&#x27;isotonic&#x27;)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-1\" type=\"checkbox\" ><label for=\"sk-estimator-id-1\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">CalibratedClassifierCV</label><div class=\"sk-toggleable__content\"><pre>CalibratedClassifierCV(base_estimator=AdaBoostClassifier(learning_rate=0.1,\n",
       "                                                         n_estimators=1500,\n",
       "                                                         random_state=42),\n",
       "                       cv=StratifiedShuffleSplit(n_splits=5, random_state=42, test_size=0.2,\n",
       "            train_size=None),\n",
       "                       method=&#x27;isotonic&#x27;)</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-2\" type=\"checkbox\" ><label for=\"sk-estimator-id-2\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">base_estimator: AdaBoostClassifier</label><div class=\"sk-toggleable__content\"><pre>AdaBoostClassifier(learning_rate=0.1, n_estimators=1500, random_state=42)</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-3\" type=\"checkbox\" ><label for=\"sk-estimator-id-3\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">AdaBoostClassifier</label><div class=\"sk-toggleable__content\"><pre>AdaBoostClassifier(learning_rate=0.1, n_estimators=1500, random_state=42)</pre></div></div></div></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "CalibratedClassifierCV(base_estimator=AdaBoostClassifier(learning_rate=0.1,\n",
       "                                                         n_estimators=1500,\n",
       "                                                         random_state=42),\n",
       "                       cv=StratifiedShuffleSplit(n_splits=5, random_state=42, test_size=0.2,\n",
       "            train_size=None),\n",
       "                       method='isotonic')"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "calibrated_best_clf_strat.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's compare the predictions:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "calibrated_best_predictions_strat = calibrated_best_clf_strat.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[295,   5],\n",
       "       [  9,  36]], dtype=int64)"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "confusion_matrix(y_test, calibrated_best_predictions_strat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[296,   4],\n",
       "       [  9,  36]], dtype=int64)"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "confusion_matrix(y_test, best_predictions_strat)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Predictions are virtually identical. Let's extract the probabilities now:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "calibrated_best_probabilities_strat = calibrated_best_clf_strat.predict_proba(X_test)[:,1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's now see the average calibrated chance that an observation is classified as reaching a living income (`true`)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.1241827393848341"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(calibrated_best_probabilities_strat)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Which now is in line with:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.11884057971014493"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(calibrated_best_predictions_strat)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 8.2 TPOT"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For completness, we do similar to the TPOT best model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "calibrated_tpot = CalibratedClassifierCV(TPOT_exported_pipeline, cv=cv_strat, method='isotonic')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "CalibratedClassifierCV(base_estimator=Pipeline(steps=[('stackingestimator',\n",
       "                                                       StackingEstimator(estimator=BernoulliNB(alpha=100.0,\n",
       "                                                                                               fit_prior=False))),\n",
       "                                                      ('rfe',\n",
       "                                                       RFE(estimator=ExtraTreesClassifier(criterion='entropy',\n",
       "                                                                                          max_features=0.8,\n",
       "                                                                                          random_state=32),\n",
       "                                                           step=0.9000000000000001)),\n",
       "                                                      ('normalizer',\n",
       "                                                       Normalizer(norm='l1')),\n",
       "                                                      ('gradientboostingclassifier',\n",
       "                                                       GradientBoostingClassifier(learning_rate=0.01,\n",
       "                                                                                  max_depth=2,\n",
       "                                                                                  max_features=0.9500000000000001,\n",
       "                                                                                  min_samples_leaf=7,\n",
       "                                                                                  min_samples_split=19,\n",
       "                                                                                  random_state=32,\n",
       "                                                                                  subsample=0.3))]),\n",
       "                       cv=StratifiedShuffleSplit(n_splits=5, random_state=42, test_size=0.2,\n",
       "            train_size=None),\n",
       "                       method='isotonic')"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "calibrated_tpot.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "calibrated_tpot_predictions = calibrated_tpot.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[298,   2],\n",
       "       [ 10,  35]], dtype=int64)"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "confusion_matrix(y_test, calibrated_tpot_predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[299,   1],\n",
       "       [ 12,  33]], dtype=int64)"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "confusion_matrix(y_test, tpot_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "calibrated_tpot_probabilities = calibrated_tpot.predict_proba(X_test)[:,1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.12618606828750847"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(calibrated_tpot_probabilities)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.1072463768115942"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(calibrated_tpot_predictions)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 8.3 Export best calibrated model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "best_calibrated_model =  calibrated_best_clf_strat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "pkl_filename = '../models/LI_simplified_model.pkl'\n",
    "pickle.dump(best_calibrated_model, open(pkl_filename, 'wb'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 9 Web app & write-ups"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 9.0 Previous non-technical write-up:\n",
    "\n",
    "See at [medium](https://medium.com/@tyszler.jobs/are-cocoa-farmers-reaching-a-living-income-f7724af574c4?sk=344c18d46a7fd402d3a137061c6ba89a)\n",
    "\n",
    "\n",
    "## 9.1 Non-technical write-up\n",
    "\n",
    "See See at [medium](https://mtyszler.medium.com/are-farmers-reaching-a-living-income-c15523a564b2)\n",
    "\n",
    "## 9.2 Technical write-up\n",
    "\n",
    "See at [medium](https://mtyszler.medium.com/are-farmers-reaching-a-living-income-the-machine-learning-technical-companion-c8e4574868da)\n",
    "\n",
    "## 9.3 Web app\n",
    "\n",
    "See it [live](https://disaster-response-livedemo.herokuapp.com/)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.10.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}