notebooks/threshold-f1-score.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "ab958018",
   "metadata": {},
   "outputs": [],
   "source": [
    "from equalityml import FAIR, discrimination_threshold, paired_ttest, binary_threshold_score\n",
    "from sklearn.linear_model import LogisticRegression, SGDClassifier\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import OneHotEncoder, StandardScaler\n",
    "from sklearn.metrics import accuracy_score, f1_score\n",
    "from sklearn.base import clone\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import warnings\n",
    "\n",
    "%matplotlib inline\n",
    "random_seed = 0"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "450e2392",
   "metadata": {},
   "source": [
    "### 1. Train a Machine Learning Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "417cb257-bbcf-4103-b0ec-9d9e13e52189",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Dataset Parameters\n",
    "target_var = 'two_year_recid'\n",
    "features = ['juv_fel_count', 'decile_score', 'juv_misd_count', 'juv_other_count', 'priors_count', 'sex', 'age']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "90e2c41e",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "data = pd.read_csv('../data/compas-scores-two-years.csv') "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "033f1b5a",
   "metadata": {},
   "outputs": [],
   "source": [
    "y = data[target_var]\n",
    "X = data[features]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "37c48622-835b-4256-966e-2d38f411a1da",
   "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>juv_fel_count</th>\n",
       "      <th>decile_score</th>\n",
       "      <th>juv_misd_count</th>\n",
       "      <th>juv_other_count</th>\n",
       "      <th>priors_count</th>\n",
       "      <th>sex</th>\n",
       "      <th>age</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Male</td>\n",
       "      <td>69</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Male</td>\n",
       "      <td>34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>Male</td>\n",
       "      <td>24</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>Male</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>Male</td>\n",
       "      <td>43</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   juv_fel_count  decile_score  juv_misd_count  juv_other_count  priors_count  \\\n",
       "0              0             1               0                0             0   \n",
       "1              0             3               0                0             0   \n",
       "2              0             4               0                1             4   \n",
       "3              0             8               1                0             1   \n",
       "4              0             1               0                0             2   \n",
       "\n",
       "    sex  age  \n",
       "0  Male   69  \n",
       "1  Male   34  \n",
       "2  Male   24  \n",
       "3  Male   23  \n",
       "4  Male   43  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "392e1749",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define which columns should be encoded\n",
    "columns_to_encode = ['sex']\n",
    "\n",
    "# Instantiate encoder\n",
    "ohe = OneHotEncoder(sparse=False, drop = 'if_binary')\n",
    "\n",
    "# Scale \n",
    "encoded_data = ohe.fit_transform(X[columns_to_encode])\n",
    "\n",
    "# Concatenate (Column-Bind) Processed Columns Back Together\n",
    "encoded_df = pd.DataFrame(encoded_data, columns = columns_to_encode)\n",
    "\n",
    "processed_X = pd.concat([X.drop(columns=columns_to_encode), encoded_df], axis=1)\n",
    "processed_X = pd.DataFrame(processed_X, columns = X.columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "22d95660-9706-4fca-9700-7675de5e2e18",
   "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>juv_fel_count</th>\n",
       "      <th>decile_score</th>\n",
       "      <th>juv_misd_count</th>\n",
       "      <th>juv_other_count</th>\n",
       "      <th>priors_count</th>\n",
       "      <th>sex</th>\n",
       "      <th>age</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>69</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>24</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>43</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   juv_fel_count  decile_score  juv_misd_count  juv_other_count  priors_count  \\\n",
       "0              0             1               0                0             0   \n",
       "1              0             3               0                0             0   \n",
       "2              0             4               0                1             4   \n",
       "3              0             8               1                0             1   \n",
       "4              0             1               0                0             2   \n",
       "\n",
       "   sex  age  \n",
       "0  1.0   69  \n",
       "1  1.0   34  \n",
       "2  1.0   24  \n",
       "3  1.0   23  \n",
       "4  1.0   43  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "processed_X.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "721e8581",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Training and testing dataset split\n",
    "X_train, X_test, y_train, y_test = train_test_split(processed_X, y, test_size=0.2, random_state=random_seed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "9c28d445",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression()"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Train a machine learning model (for example LogisticRegression)\n",
    "orig_ml_model = LogisticRegression()\n",
    "orig_ml_model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0ced551e-100f-486c-9890-d01e2cd47d92",
   "metadata": {},
   "source": [
    "### 2. Optimal Discrimination Threshold for F1 score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "69cfde39-494f-455c-9c78-7d517caf0dfe",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10/10 [00:01<00:00,  6.22it/s]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Discrimination Threshold = 0.364\n"
     ]
    }
   ],
   "source": [
    "decision_maker = ['f1', 'max']\n",
    "scoring = \"f1\"\n",
    "# Discrimination Threshold\n",
    "orig_threshold = discrimination_threshold(orig_ml_model, \n",
    "                                         processed_X,\n",
    "                                         y,\n",
    "                                         decision_maker=decision_maker,\n",
    "                                         metrics=[decision_maker[0]], \n",
    "                                         show=True,\n",
    "                                         random_seed=random_seed)\n",
    "print(f\"Discrimination Threshold = {orig_threshold:.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "4304469c-4f45-4e5d-800d-853fcf7bb81f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Original F1 score = 0.683\n"
     ]
    }
   ],
   "source": [
    "orig_f1_score = binary_threshold_score(orig_ml_model, \n",
    "                                        X_test, \n",
    "                                        y_test,\n",
    "                                        scoring=scoring,  \n",
    "                                        threshold=orig_threshold)\n",
    "print(f\"Original F1 score = {orig_f1_score:.3f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7089a26b-0199-4210-b1fb-9eb4a71f1495",
   "metadata": {},
   "source": [
    "### 3. Fairness Metric Assessment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "99c9479d",
   "metadata": {},
   "outputs": [],
   "source": [
    "testing_data = pd.concat([X_test, y_test], axis=1)\n",
    "training_data = pd.concat([X_train, y_train], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "7e0e58a2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Instantiate FAIR object\n",
    "fair_object = FAIR(ml_model=orig_ml_model,\n",
    "                   training_data=training_data,\n",
    "                   testing_data=testing_data, \n",
    "                   target_variable=target_var,\n",
    "                   protected_variable=\"sex\", \n",
    "                   privileged_class=1.0,\n",
    "                   threshold=orig_threshold,\n",
    "                   random_seed=random_seed) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "2c2babfc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Original Fairness Metric = 0.633\n"
     ]
    }
   ],
   "source": [
    "metric_name='statistical_parity_ratio'\n",
    "orig_fairness_metric_score = fair_object.fairness_metric(metric_name)\n",
    "print(f\"Original Fairness Metric = {orig_fairness_metric_score:.3f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "12be4fed-1ee1-4173-b2a8-fa233299cadd",
   "metadata": {
    "tags": []
   },
   "source": [
    "### 4. Compare Performance of Each Mitigation Method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "96389970",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Available bias mitigation methods for 'statistical_parity_ratio' are:\n",
      "1 'disparate-impact-remover'\n",
      "2 'resampling'\n",
      "3 'resampling-preferential'\n",
      "4 'reweighing'\n"
     ]
    }
   ],
   "source": [
    "# Print available bias mitigation methods for statistical_parity_ratio\n",
    "fair_object.print_bias_mitigation_methods(metric_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "a0b07834-61b4-4296-9226-a6aa549f9f8b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                                f1  statistical_parity_ratio  \\\n",
      "reference                 0.683181                  0.632963   \n",
      "disparate-impact-remover  0.685526                  0.624395   \n",
      "resampling                0.687693                  0.975532   \n",
      "resampling-preferential   0.684505                  0.874168   \n",
      "reweighing                0.688545                  0.976120   \n",
      "\n",
      "                          discrimination_threshold  \n",
      "reference                                 0.363636  \n",
      "disparate-impact-remover                  0.363636  \n",
      "resampling                                0.333333  \n",
      "resampling-preferential                   0.333333  \n",
      "reweighing                                0.333333  \n"
     ]
    }
   ],
   "source": [
    "df = fair_object.compare_mitigation_methods(metric_name=metric_name, \n",
    "                                            scoring=scoring,                                             \n",
    "                                            show=False,\n",
    "                                            compute_discrimination_threshold=True,\n",
    "                                            decision_maker=decision_maker)\n",
    "print(df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "50f42439",
   "metadata": {},
   "source": [
    "### 5. Apply Resampling Mitigation and Re-evaluate F1 Score and Fairness Metric"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "c52cefb8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression()"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mitigation_method = \"resampling\"\n",
    "mitigation = fair_object.bias_mitigation(mitigation_method = mitigation_method)\n",
    "\n",
    "# extract modified data \n",
    "X_train = mitigation['training_data'][features]\n",
    "# drop y \n",
    "y_train = mitigation['training_data'][target_var]\n",
    "\n",
    "# Train the second machine learning model using the mitigated dataset\n",
    "ml_model = clone(orig_ml_model)\n",
    "ml_model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "6c866ae1-f21d-4131-af51-b2c5dfe0f698",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Discrimination Threshold = 0.333\n"
     ]
    }
   ],
   "source": [
    "# We can also find the optimal discrimination threshold for this new model\n",
    "threshold = discrimination_threshold(ml_model, \n",
    "                                     X_test,\n",
    "                                     y_test,\n",
    "                                     decision_maker=decision_maker,\n",
    "                                     metrics=[decision_maker[0]], \n",
    "                                     show=True,\n",
    "                                     model_training=False,\n",
    "                                     random_seed=random_seed)\n",
    "print(f\"Discrimination Threshold = {threshold:.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "67ac3dbe-f8e1-4b47-bb97-5c9ab6b1e2b3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mitigated F1 score = 0.689\n"
     ]
    }
   ],
   "source": [
    "f1_score = binary_threshold_score(ml_model, \n",
    "                                  X_test, \n",
    "                                  y_test,\n",
    "                                  scoring=scoring, \n",
    "                                  threshold=threshold)\n",
    "print(f\"Mitigated F1 score = {f1_score:.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "21ef4274-d148-4ac6-9b12-c74c88d272e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Instantiate FAIR object\n",
    "fair_object = FAIR(ml_model=ml_model,\n",
    "                   training_data=training_data,\n",
    "                   testing_data=testing_data, \n",
    "                   target_variable=target_var,\n",
    "                   protected_variable=\"sex\", \n",
    "                   privileged_class=1.0,\n",
    "                   threshold=threshold,\n",
    "                   random_seed=random_seed) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "e7a2d35a-14cd-409f-a16f-ef0d0b7d286c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mitigated Fairness Metric = 0.971\n"
     ]
    }
   ],
   "source": [
    "fairness_metric_score = fair_object.fairness_metric(metric_name)\n",
    "print(f\"Mitigated Fairness Metric = {fairness_metric_score:.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "8b795994-5432-46c0-9a3a-24ee5c630535",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Original F1 score = 0.683 and Mitigated F1 score = 0.689\n",
      "Orginal Fairness Score = 0.633 and Mitigated Fairness Score = 0.971\n"
     ]
    }
   ],
   "source": [
    "print(f\"Original F1 score = {orig_f1_score:.3f} and Mitigated F1 score = {f1_score:.3f}\")\n",
    "print(f\"Orginal Fairness Score = {orig_fairness_metric_score:.3f} and Mitigated Fairness Score = {fairness_metric_score:.3f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "619cc2f7-5d99-4f61-bc56-ac54c23f167b",
   "metadata": {},
   "source": [
    "### 6. Paired t Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "88ca3648-ce64-4299-bf45-87cb62b96b36",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 5/5 [00:02<00:00,  1.82it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5x2cv paired t-test using fairness metric. (chi2, p) = (-7.5726397407027415, 0.0006370370656437617)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# Paired t test between original Machine Learning model and \"applying the mitigation method\".\n",
    "# Fairness metric comparison\n",
    "results = paired_ttest(orig_ml_model,\n",
    "                        processed_X,\n",
    "                        y,\n",
    "                        method=\"5x2cv\",\n",
    "                        threshold=orig_threshold,\n",
    "                        fair_object=fair_object,\n",
    "                        mitigation_method=mitigation_method,\n",
    "                        scoring=metric_name,\n",
    "                        random_seed=random_seed)\n",
    "print(\"5x2cv paired t-test using fairness metric. (chi2, p) =\", results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "36a9f9fb-567e-4724-babb-7042a2b7bf62",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 5/5 [00:05<00:00,  1.14s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5x2cv paired t-test using fairness metric. (chi2, p) = (-4.239732269233114, 0.008171698334078475)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# Paired t test between original Machine Learning model and \"applying the mitigation method\".\n",
    "# Fairness metric comparison computing the discrimination threshold\n",
    "results = paired_ttest(orig_ml_model,\n",
    "                        processed_X,\n",
    "                        y,\n",
    "                        method=\"5x2cv\",\n",
    "                        fair_object=fair_object,\n",
    "                        mitigation_method=mitigation_method,\n",
    "                        scoring=metric_name,\n",
    "                        compute_discrimination_threshold=True,\n",
    "                        decision_maker=decision_maker,\n",
    "                        random_seed=random_seed)\n",
    "print(\"5x2cv paired t-test using fairness metric. (chi2, p) =\", results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "dfc306a6-14a8-4caf-a0b0-de3a145f7f62",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 5/5 [00:01<00:00,  4.39it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5x2cv paired t-test using F1 metric. (chi2, p) = (2.36023417517857, 0.06473221235261217)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# Paired t test between original Machine Learning model and \"applying the mitigation method\".\n",
    "# F1 score comparison\n",
    "results = paired_ttest(orig_ml_model,\n",
    "                        processed_X,\n",
    "                        y,\n",
    "                        method=\"5x2cv\",\n",
    "                        threshold=orig_threshold,\n",
    "                        fair_object=fair_object,\n",
    "                        mitigation_method=mitigation_method,\n",
    "                        scoring=scoring,\n",
    "                        random_seed=random_seed)\n",
    "print(\"5x2cv paired t-test using F1 metric. (chi2, p) =\", results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "7afc0cf3-2651-4256-9f44-bdc220d47b5f",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 5/5 [00:04<00:00,  1.06it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5x2cv paired t-test using F1 metric. (chi2, p) = (0.7378951559332236, 0.4937314596908805)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# Paired t test between original Machine Learning model and \"applying the mitigation method\".\n",
    "# F1 score comparison computing the discrimination threshold\n",
    "results = paired_ttest(orig_ml_model,\n",
    "                        processed_X,\n",
    "                        y,\n",
    "                        method=\"5x2cv\",\n",
    "                        fair_object=fair_object,\n",
    "                        mitigation_method=mitigation_method,\n",
    "                        scoring=scoring,\n",
    "                        compute_discrimination_threshold=True,\n",
    "                        decision_maker=decision_maker,\n",
    "                        random_seed=random_seed)\n",
    "print(\"5x2cv paired t-test using F1 metric. (chi2, p) =\", results)"
   ]
  }
 ],
 "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": "3.8.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}