main.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Bias Mitigation using regularisation and reweighing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Init"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from dataset import Dataset\n",
    "from models import LogisticRegressionClassifier\n",
    "from analysis import Analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Adult Dataset initialized\n",
      "German Dataset initialized\n"
     ]
    }
   ],
   "source": [
    "adult_ds = Dataset(dataset_name=\"adult\")\n",
    "german_ds = Dataset(dataset_name=\"german\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "C_values = [0.0001, 0.001, 0.01, 0.1, 1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Adult Income Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "selected_dataset = adult_ds\n",
    "X_train, y_train, X_test, y_test = selected_dataset.get_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Running through Solvers and Penalties"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "models = [{'solver': \"saga\", \"params\":{'penalty': \"l2\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"saga\", \"params\":{'penalty': \"l1\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"saga\", \"params\":{'penalty': \"elasticnet\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4, 'l1_ratio': 0.5}},\n",
    "          {'solver': \"saga\", \"params\":{'penalty': \"none\"}},\n",
    "          {'solver': \"sag\", \"params\":{'penalty': \"l2\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"sag\", \"params\":{'penalty': \"none\"}},\n",
    "          {'solver': \"liblinear\", \"params\":{'penalty': \"l1\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"liblinear\", \"params\":{'penalty': \"l2\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"lbfgs\", \"params\":{'penalty': \"l2\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"lbfgs\", \"params\":{'penalty': \"none\"}},\n",
    "          {'solver': \"newton-cg\", \"params\":{'penalty': \"l2\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"newton-cg\", \"params\":{'penalty': \"none\"}}]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running model: LogisticRegressionClassifier solver saga with l2 regularization\n",
      "Running model: LogisticRegressionClassifier solver saga with l1 regularization\n",
      "Running model: LogisticRegressionClassifier solver saga with elasticnet regularization\n",
      "Running model: LogisticRegressionClassifier solver saga with none regularization\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "The max_iter was reached which means the coef_ did not converge\n",
      "The max_iter was reached which means the coef_ did not converge\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running model: LogisticRegressionClassifier solver sag with l2 regularization\n",
      "Running model: LogisticRegressionClassifier solver sag with none regularization\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "The max_iter was reached which means the coef_ did not converge\n",
      "The max_iter was reached which means the coef_ did not converge\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running model: LogisticRegressionClassifier solver liblinear with l1 regularization\n",
      "Running model: LogisticRegressionClassifier solver liblinear with l2 regularization\n",
      "Running model: LogisticRegressionClassifier solver lbfgs with l2 regularization\n",
      "Running model: LogisticRegressionClassifier solver lbfgs with none regularization\n",
      "Running model: LogisticRegressionClassifier solver newton-cg with l2 regularization\n",
      "Running model: LogisticRegressionClassifier solver newton-cg with none regularization\n"
     ]
    }
   ],
   "source": [
    "results = []\n",
    "\n",
    "for model in models:\n",
    "    clf = LogisticRegressionClassifier(verbose=False, dataset=selected_dataset, solver=model['solver'], params=model['params'], privileged_groups=selected_dataset.privileged_groups, unprivileged_groups=selected_dataset.unprivileged_groups)\n",
    "    print(\"Running model: \" + str(clf.__class__.__name__) + \" solver \" + model['solver'] + \" with \" + str(model['params']['penalty']) + \" regularization\")\n",
    "    results.append((model['solver'] + \"_\" + str(model['params']['penalty']), *clf.run(selected_dataset)))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Solver-Penalty Test\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>std</th>\n",
       "      <th>fairness_metric</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>saga_l2</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.00434</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>saga_l1</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.00434</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>saga_elasticnet</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.00434</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>saga_none</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.00434</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>sag_l2</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.00434</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>sag_none</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.00434</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>liblinear_l1</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.00434</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>liblinear_l2</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.00434</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>lbfgs_l2</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.00434</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>lbfgs_none</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.00434</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>newton-cg_l2</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.00434</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>newton-cg_none</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.00434</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              model  accuracy      std  fairness_metric\n",
       "0           saga_l2  0.804001  0.00434        -0.464784\n",
       "1           saga_l1  0.804001  0.00434        -0.464784\n",
       "2   saga_elasticnet  0.804001  0.00434        -0.464784\n",
       "3         saga_none  0.804001  0.00434        -0.464784\n",
       "4            sag_l2  0.804001  0.00434        -0.464784\n",
       "5          sag_none  0.804001  0.00434        -0.464784\n",
       "6      liblinear_l1  0.804001  0.00434        -0.464784\n",
       "7      liblinear_l2  0.804001  0.00434        -0.464784\n",
       "8          lbfgs_l2  0.804001  0.00434        -0.464784\n",
       "9        lbfgs_none  0.804001  0.00434        -0.464784\n",
       "10     newton-cg_l2  0.804001  0.00434        -0.464784\n",
       "11   newton-cg_none  0.804001  0.00434        -0.464784"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"Solver-Penalty Test\")\n",
    "\n",
    "# display results\n",
    "df = pd.DataFrame(results, columns=['model', 'accuracy', 'std', 'fairness_metric'])\n",
    "df = df.sort_values(by=['accuracy'], ascending=False) # sort by accuracy\n",
    "\n",
    "# show\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Task 1: Machine Learning Based Model\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Hyperparameter search over C (Lambda)\n",
    "\n",
    "- Solver: liblinear\n",
    "- Dataset: Adult\n",
    "- C: 0.0001 to 0.1\n",
    "- Penalty: l2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy: 3\n",
      "Best fairness: 0\n",
      "Best accuracy+fairness: None\n"
     ]
    }
   ],
   "source": [
    "analyser = Analysis(dataset=selected_dataset, C_values=C_values, solver=\"liblinear\", params={\"penalty\": \"l2\"}, mitigated=False)\n",
    "analysis = analyser.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Performance measures over hyperparameter search:\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>accuracy_std</th>\n",
       "      <th>fairness</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0001</td>\n",
       "      <td>0.797216</td>\n",
       "      <td>0.005006</td>\n",
       "      <td>-0.242389</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0010</td>\n",
       "      <td>0.803387</td>\n",
       "      <td>0.004469</td>\n",
       "      <td>-0.454879</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0100</td>\n",
       "      <td>0.803914</td>\n",
       "      <td>0.004465</td>\n",
       "      <td>-0.462947</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.1000</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.004340</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.0000</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.004340</td>\n",
       "      <td>-0.464784</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    model  accuracy  accuracy_std  fairness\n",
       "0  0.0001  0.797216      0.005006 -0.242389\n",
       "1  0.0010  0.803387      0.004469 -0.454879\n",
       "2  0.0100  0.803914      0.004465 -0.462947\n",
       "3  0.1000  0.804001      0.004340 -0.464784\n",
       "4  1.0000  0.804001      0.004340 -0.464784"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"Performance measures over hyperparameter search:\")\n",
    "analysis['data_frame']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 576x396 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "analyser.plot_results()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Accuracy and Fairness on test data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__[Model 1]: Best Accuracy__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Best accuracy] Accuracy on test set: 0.8042039172865625\n",
      "[Best accuracy] Fairness (True Positive Rate Difference) on test set: -0.4414141414141414\n"
     ]
    }
   ],
   "source": [
    "# Best accuracy\n",
    "test_accuracy = analysis['best_accuracy_model'].get_accuracy(X_test, y_test)\n",
    "test_fairness = analysis['best_accuracy_model'].get_fairness(selected_dataset.dataset_orig_test)\n",
    "\n",
    "print(\"[Best accuracy] Accuracy on test set: \" + str(test_accuracy))\n",
    "print(\"[Best accuracy] Fairness (True Positive Rate Difference) on test set: \" + str(test_fairness['eq_opp_diff']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__[Model 2]: Best Fairness__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Best fairness] Accuracy on test set: 0.7984030573943902\n",
      "[Best fairness] Fairness (True Positive Rate Difference) on test set: -0.2472160630697216\n"
     ]
    }
   ],
   "source": [
    "# Best fairness\n",
    "test_accuracy = analysis['best_fairness_model'].get_accuracy(X_test, y_test)\n",
    "test_fairness = analysis['best_fairness_model'].get_fairness(selected_dataset.dataset_orig_test)\n",
    "\n",
    "print(\"[Best fairness] Accuracy on test set: \" + str(test_accuracy))\n",
    "print(\"[Best fairness] Fairness (True Positive Rate Difference) on test set: \" + str(test_fairness['eq_opp_diff']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Task 2: Fairness Based Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Bias mitigation using reweighing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mitigation: Reweighing complete\n"
     ]
    }
   ],
   "source": [
    "selected_dataset.dataset_orig_train = selected_dataset.mitigate_bias(method=\"reweighing\", dataset=selected_dataset.dataset_orig_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy: 3\n",
      "Best fairness: 0\n",
      "Best accuracy+fairness: None\n"
     ]
    }
   ],
   "source": [
    "analyser = Analysis(dataset=selected_dataset, C_values=C_values, solver=\"liblinear\", params={\"penalty\": \"l2\"}, mitigated=True)\n",
    "analysis = analyser.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Results after mitigation:\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>accuracy_std</th>\n",
       "      <th>fairness</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0001</td>\n",
       "      <td>0.797216</td>\n",
       "      <td>0.005006</td>\n",
       "      <td>-0.009967</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0010</td>\n",
       "      <td>0.803387</td>\n",
       "      <td>0.004469</td>\n",
       "      <td>-0.009967</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0100</td>\n",
       "      <td>0.803914</td>\n",
       "      <td>0.004465</td>\n",
       "      <td>-0.009967</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.1000</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.004340</td>\n",
       "      <td>-0.009967</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.0000</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.004340</td>\n",
       "      <td>-0.009967</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    model  accuracy  accuracy_std  fairness\n",
       "0  0.0001  0.797216      0.005006 -0.009967\n",
       "1  0.0010  0.803387      0.004469 -0.009967\n",
       "2  0.0100  0.803914      0.004465 -0.009967\n",
       "3  0.1000  0.804001      0.004340 -0.009967\n",
       "4  1.0000  0.804001      0.004340 -0.009967"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"Results after mitigation:\")\n",
    "analysis['data_frame']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Accuracy and Fairness on test data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__[Model 3]: Best Accuracy__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Best accuracy] Accuracy on test set: 0.7905548351873336\n",
      "[Best accuracy] Fairness (True Positive Rate Difference) on test set: 0.03513180586351322\n"
     ]
    }
   ],
   "source": [
    "# Best accuracy\n",
    "test_accuracy = analysis['best_accuracy_model'].get_accuracy(X_test, y_test)\n",
    "test_fairness = analysis['best_accuracy_model'].get_fairness(selected_dataset.dataset_orig_test)\n",
    "\n",
    "print(\"[Best accuracy] Accuracy on test set: \" + str(test_accuracy))\n",
    "print(\"[Best accuracy] Fairness (True Positive Rate Difference) on test set: \" + str(test_fairness['eq_opp_diff']))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__[Model 4]: Best Fairness__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Best fairness] Accuracy on test set: 0.7905548351873336\n",
      "[Best fairness] Fairness (True Positive Rate Difference) on test set: 0.03513180586351322\n"
     ]
    }
   ],
   "source": [
    "# Best fairness\n",
    "test_accuracy = analysis['best_fairness_model'].get_accuracy(X_test, y_test)\n",
    "test_fairness = analysis['best_fairness_model'].get_fairness(selected_dataset.dataset_orig_test)\n",
    "\n",
    "print(\"[Best fairness] Accuracy on test set: \" + str(test_accuracy))\n",
    "print(\"[Best fairness] Fairness (True Positive Rate Difference) on test set: \" + str(test_fairness['eq_opp_diff']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 576x396 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "analyser.plot_results()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Task 3: Accuracy + Fairness Based Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Machine-Learning based model using accuracy+fairness criterion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy: 3\n",
      "Best fairness: 0\n",
      "Best accuracy+fairness: 0\n"
     ]
    }
   ],
   "source": [
    "analyser = Analysis(dataset=selected_dataset, C_values=C_values, solver=\"liblinear\", params={\"penalty\": \"l2\"}, mitigated=False, include_delta=True)\n",
    "analysis = analyser.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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>model</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>accuracy_std</th>\n",
       "      <th>fairness</th>\n",
       "      <th>accuracy+fairness</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0001</td>\n",
       "      <td>0.797216</td>\n",
       "      <td>0.005006</td>\n",
       "      <td>-0.242389</td>\n",
       "      <td>1.039605</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0010</td>\n",
       "      <td>0.803387</td>\n",
       "      <td>0.004469</td>\n",
       "      <td>-0.454879</td>\n",
       "      <td>1.258646</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0100</td>\n",
       "      <td>0.803914</td>\n",
       "      <td>0.004465</td>\n",
       "      <td>-0.462947</td>\n",
       "      <td>1.267241</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.1000</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.004340</td>\n",
       "      <td>-0.464784</td>\n",
       "      <td>1.269078</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.0000</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.004340</td>\n",
       "      <td>-0.464784</td>\n",
       "      <td>1.269078</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    model  accuracy  accuracy_std  fairness  accuracy+fairness\n",
       "0  0.0001  0.797216      0.005006 -0.242389           1.039605\n",
       "1  0.0010  0.803387      0.004469 -0.454879           1.258646\n",
       "2  0.0100  0.803914      0.004465 -0.462947           1.267241\n",
       "3  0.1000  0.804001      0.004340 -0.464784           1.269078\n",
       "4  1.0000  0.804001      0.004340 -0.464784           1.269078"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "analysis['data_frame']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 576x396 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "analyser.plot_results(type=\"accuracy+fairness\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Accuracy and Fairness on test data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__[Model 5]: Best Accuracy__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test set: 0.7984030573943902\n",
      "Fairness (True Positive Rate Difference) on test set: -0.2472160630697216\n"
     ]
    }
   ],
   "source": [
    "# Best accuracy+fairness\n",
    "test_accuracy = analysis['best_accuracy_fairness_model'].get_accuracy(X_test, y_test)\n",
    "test_fairness = analysis['best_accuracy_fairness_model'].get_fairness(selected_dataset.dataset_orig_test)\n",
    "\n",
    "print(\"Accuracy on test set: \" + str(test_accuracy))\n",
    "print(\"Fairness (True Positive Rate Difference) on test set: \" + str(test_fairness['eq_opp_diff']))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Fairness based model using accuracy+fairness criterion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy: 3\n",
      "Best fairness: 0\n",
      "Best accuracy+fairness: 0\n"
     ]
    }
   ],
   "source": [
    "analyser = Analysis(dataset=selected_dataset, C_values=C_values, solver=\"liblinear\", params={\"penalty\": \"l2\"}, mitigated=True, include_delta=True)\n",
    "analysis = analyser.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "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>model</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>accuracy_std</th>\n",
       "      <th>fairness</th>\n",
       "      <th>accuracy+fairness</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0001</td>\n",
       "      <td>0.797216</td>\n",
       "      <td>0.005006</td>\n",
       "      <td>-0.009967</td>\n",
       "      <td>0.799666</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0010</td>\n",
       "      <td>0.803387</td>\n",
       "      <td>0.004469</td>\n",
       "      <td>-0.009967</td>\n",
       "      <td>0.799666</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0100</td>\n",
       "      <td>0.803914</td>\n",
       "      <td>0.004465</td>\n",
       "      <td>-0.009967</td>\n",
       "      <td>0.799666</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.1000</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.004340</td>\n",
       "      <td>-0.009967</td>\n",
       "      <td>0.799666</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.0000</td>\n",
       "      <td>0.804001</td>\n",
       "      <td>0.004340</td>\n",
       "      <td>-0.009967</td>\n",
       "      <td>0.799666</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    model  accuracy  accuracy_std  fairness  accuracy+fairness\n",
       "0  0.0001  0.797216      0.005006 -0.009967           0.799666\n",
       "1  0.0010  0.803387      0.004469 -0.009967           0.799666\n",
       "2  0.0100  0.803914      0.004465 -0.009967           0.799666\n",
       "3  0.1000  0.804001      0.004340 -0.009967           0.799666\n",
       "4  1.0000  0.804001      0.004340 -0.009967           0.799666"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "analysis['data_frame']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 576x396 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "analyser.plot_results(type=\"accuracy+fairness\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Accuracy and Fairness on test data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__[Model 6]: Best Fairness__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test set: 0.7905548351873336\n",
      "Fairness (True Positive Rate Difference) on test set: 0.03513180586351322\n"
     ]
    }
   ],
   "source": [
    "# Best accuracy+fairness\n",
    "test_accuracy = analysis['best_accuracy_fairness_model'].get_accuracy(X_test, y_test)\n",
    "test_fairness = analysis['best_accuracy_fairness_model'].get_fairness(selected_dataset.dataset_orig_test)\n",
    "\n",
    "print(\"Accuracy on test set: \" + str(test_accuracy))\n",
    "print(\"Fairness (True Positive Rate Difference) on test set: \" + str(test_fairness['eq_opp_diff']))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "______________________________________________________________________________________________________________________________________________________________"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "______________________________________________________________________________________________________________________________________________________________"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "______________________________________________________________________________________________________________________________________________________________"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# German Census Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "selected_dataset = german_ds\n",
    "X_train, y_train, X_test, y_test = selected_dataset.get_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Running through Solvers and Penalties"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "models = [{'solver': \"saga\", \"params\":{'penalty': \"l2\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"saga\", \"params\":{'penalty': \"l1\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"saga\", \"params\":{'penalty': \"elasticnet\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4, 'l1_ratio': 0.5}},\n",
    "          {'solver': \"saga\", \"params\":{'penalty': \"none\"}},\n",
    "          {'solver': \"sag\", \"params\":{'penalty': \"l2\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"sag\", \"params\":{'penalty': \"none\"}},\n",
    "          {'solver': \"liblinear\", \"params\":{'penalty': \"l1\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"liblinear\", \"params\":{'penalty': \"l2\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"lbfgs\", \"params\":{'penalty': \"l2\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"lbfgs\", \"params\":{'penalty': \"none\"}},\n",
    "          {'solver': \"newton-cg\", \"params\":{'penalty': \"l2\", 'C': 1, 'max_iter': 1000, 'tol': 1e-4}},\n",
    "          {'solver': \"newton-cg\", \"params\":{'penalty': \"none\"}}]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running model: LogisticRegressionClassifier solver saga with l2 regularization\n",
      "Running model: LogisticRegressionClassifier solver saga with l1 regularization\n",
      "Running model: LogisticRegressionClassifier solver saga with elasticnet regularization\n",
      "Running model: LogisticRegressionClassifier solver saga with none regularization\n",
      "Running model: LogisticRegressionClassifier solver sag with l2 regularization\n",
      "Running model: LogisticRegressionClassifier solver sag with none regularization\n",
      "Running model: LogisticRegressionClassifier solver liblinear with l1 regularization\n",
      "Running model: LogisticRegressionClassifier solver liblinear with l2 regularization\n",
      "Running model: LogisticRegressionClassifier solver lbfgs with l2 regularization\n",
      "Running model: LogisticRegressionClassifier solver lbfgs with none regularization\n",
      "Running model: LogisticRegressionClassifier solver newton-cg with l2 regularization\n",
      "Running model: LogisticRegressionClassifier solver newton-cg with none regularization\n"
     ]
    }
   ],
   "source": [
    "results = []\n",
    "\n",
    "for model in models:\n",
    "    clf = LogisticRegressionClassifier(verbose=False, dataset=selected_dataset, solver=model['solver'], params=model['params'], privileged_groups=selected_dataset.privileged_groups, unprivileged_groups=selected_dataset.unprivileged_groups)\n",
    "    print(\"Running model: \" + str(clf.__class__.__name__) + \" solver \" + model['solver'] + \" with \" + str(model['params']['penalty']) + \" regularization\")\n",
    "    results.append((model['solver'] + \"_\" + str(model['params']['penalty']), *clf.run(selected_dataset)))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "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>model</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>std</th>\n",
       "      <th>fairness_metric</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>saga_l2</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>-0.133386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>saga_elasticnet</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>-0.133386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>saga_none</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>-0.133386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>sag_l2</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>-0.133386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>sag_none</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>-0.133386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>liblinear_l2</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>-0.133386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>lbfgs_l2</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>-0.133386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>lbfgs_none</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>-0.133386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>newton-cg_l2</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>-0.133386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>newton-cg_none</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>-0.133386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>saga_l1</td>\n",
       "      <td>0.690000</td>\n",
       "      <td>0.026573</td>\n",
       "      <td>-0.144980</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>liblinear_l1</td>\n",
       "      <td>0.690000</td>\n",
       "      <td>0.026573</td>\n",
       "      <td>-0.144980</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              model  accuracy       std  fairness_metric\n",
       "0           saga_l2  0.692857  0.031623        -0.133386\n",
       "2   saga_elasticnet  0.692857  0.031623        -0.133386\n",
       "3         saga_none  0.692857  0.031623        -0.133386\n",
       "4            sag_l2  0.692857  0.031623        -0.133386\n",
       "5          sag_none  0.692857  0.031623        -0.133386\n",
       "7      liblinear_l2  0.692857  0.031623        -0.133386\n",
       "8          lbfgs_l2  0.692857  0.031623        -0.133386\n",
       "9        lbfgs_none  0.692857  0.031623        -0.133386\n",
       "10     newton-cg_l2  0.692857  0.031623        -0.133386\n",
       "11   newton-cg_none  0.692857  0.031623        -0.133386\n",
       "1           saga_l1  0.690000  0.026573        -0.144980\n",
       "6      liblinear_l1  0.690000  0.026573        -0.144980"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "# display results\n",
    "df = pd.DataFrame(results, columns=['model', 'accuracy', 'std', 'fairness_metric'])\n",
    "df = df.sort_values(by=['accuracy'], ascending=False) # sort by accuracy\n",
    "\n",
    "# show\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Task 1: Machine Learning Based Model\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Hyperparameter search over C (Lambda)\n",
    "\n",
    "- Solver: liblinear\n",
    "- Dataset: \"Adult | German\"\n",
    "- C: 0.0001 to 0.1\n",
    "- Penalty: l2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy: 3\n",
      "Best fairness: 4\n",
      "Best accuracy+fairness: None\n"
     ]
    }
   ],
   "source": [
    "analyser = Analysis(dataset=selected_dataset, C_values=C_values, solver=\"liblinear\", params={\"penalty\": \"l2\"}, mitigated=False)\n",
    "analysis = analyser.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "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>model</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>accuracy_std</th>\n",
       "      <th>fairness</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0001</td>\n",
       "      <td>0.655714</td>\n",
       "      <td>0.023647</td>\n",
       "      <td>-0.157834</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0010</td>\n",
       "      <td>0.660000</td>\n",
       "      <td>0.024578</td>\n",
       "      <td>-0.238396</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0100</td>\n",
       "      <td>0.687143</td>\n",
       "      <td>0.029067</td>\n",
       "      <td>-0.294951</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.1000</td>\n",
       "      <td>0.695714</td>\n",
       "      <td>0.029485</td>\n",
       "      <td>-0.139268</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.0000</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>-0.133386</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    model  accuracy  accuracy_std  fairness\n",
       "0  0.0001  0.655714      0.023647 -0.157834\n",
       "1  0.0010  0.660000      0.024578 -0.238396\n",
       "2  0.0100  0.687143      0.029067 -0.294951\n",
       "3  0.1000  0.695714      0.029485 -0.139268\n",
       "4  1.0000  0.692857      0.031623 -0.133386"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "analysis['data_frame']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 576x396 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "analyser.plot_results()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Accuracy and Fairness on test data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__[Model 1]: Best Accuracy__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Best accuracy] Accuracy on test set: 0.67\n",
      "[Best accuracy] Fairness (True Positive Rate Difference) on test set: -0.26812499999999995\n"
     ]
    }
   ],
   "source": [
    "# Best accuracy\n",
    "test_accuracy = analysis['best_accuracy_model'].get_accuracy(X_test, y_test)\n",
    "test_fairness = analysis['best_accuracy_model'].get_fairness(selected_dataset.dataset_orig_test)\n",
    "\n",
    "print(\"[Best accuracy] Accuracy on test set: \" + str(test_accuracy))\n",
    "print(\"[Best accuracy] Fairness (True Positive Rate Difference) on test set: \" + str(test_fairness['eq_opp_diff']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__[Model 2]: Best Fairness__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Best fairness] Accuracy on test set: 0.7\n",
      "[Best fairness] Fairness (True Positive Rate Difference) on test set: -0.06499999999999995\n"
     ]
    }
   ],
   "source": [
    "# Best fairness\n",
    "test_accuracy = analysis['best_fairness_model'].get_accuracy(X_test, y_test)\n",
    "test_fairness = analysis['best_fairness_model'].get_fairness(selected_dataset.dataset_orig_test)\n",
    "\n",
    "print(\"[Best fairness] Accuracy on test set: \" + str(test_accuracy))\n",
    "print(\"[Best fairness] Fairness (True Positive Rate Difference) on test set: \" + str(test_fairness['eq_opp_diff']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Task 2: Fairness Based Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Bias mitigation using reweighing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mitigation: Reweighing complete\n"
     ]
    }
   ],
   "source": [
    "selected_dataset.dataset_orig_train = selected_dataset.mitigate_bias(method=\"reweighing\", dataset=selected_dataset.dataset_orig_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy: 3\n",
      "Best fairness: 3\n",
      "Best accuracy+fairness: None\n"
     ]
    }
   ],
   "source": [
    "analyser = Analysis(dataset=selected_dataset, C_values=C_values, solver=\"liblinear\", params={\"penalty\": \"l2\"}, mitigated=True)\n",
    "analysis = analyser.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Results after mitigation:\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>accuracy_std</th>\n",
       "      <th>fairness</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0001</td>\n",
       "      <td>0.655714</td>\n",
       "      <td>0.023647</td>\n",
       "      <td>-0.139876</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0010</td>\n",
       "      <td>0.660000</td>\n",
       "      <td>0.024578</td>\n",
       "      <td>-0.103993</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0100</td>\n",
       "      <td>0.687143</td>\n",
       "      <td>0.029067</td>\n",
       "      <td>-0.044911</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.1000</td>\n",
       "      <td>0.695714</td>\n",
       "      <td>0.029485</td>\n",
       "      <td>0.003592</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.0000</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>0.047363</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    model  accuracy  accuracy_std  fairness\n",
       "0  0.0001  0.655714      0.023647 -0.139876\n",
       "1  0.0010  0.660000      0.024578 -0.103993\n",
       "2  0.0100  0.687143      0.029067 -0.044911\n",
       "3  0.1000  0.695714      0.029485  0.003592\n",
       "4  1.0000  0.692857      0.031623  0.047363"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"Results after mitigation:\")\n",
    "analysis['data_frame']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Accuracy and Fairness on test data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__[Model 3]: Best Accuracy__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Best accuracy] Accuracy on test set: 0.7066666666666667\n",
      "[Best accuracy] Fairness (True Positive Rate Difference) on test set: -0.018124999999999947\n"
     ]
    }
   ],
   "source": [
    "# Best accuracy\n",
    "test_accuracy = analysis['best_accuracy_model'].get_accuracy(X_test, y_test)\n",
    "test_fairness = analysis['best_accuracy_model'].get_fairness(selected_dataset.dataset_orig_test)\n",
    "\n",
    "print(\"[Best accuracy] Accuracy on test set: \" + str(test_accuracy))\n",
    "print(\"[Best accuracy] Fairness (True Positive Rate Difference) on test set: \" + str(test_fairness['eq_opp_diff']))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__[Model 4]: Best Fairness__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Best fairness] Accuracy on test set: 0.7066666666666667\n",
      "[Best fairness] Fairness (True Positive Rate Difference) on test set: -0.018124999999999947\n"
     ]
    }
   ],
   "source": [
    "# Best fairness\n",
    "test_accuracy = analysis['best_fairness_model'].get_accuracy(X_test, y_test)\n",
    "test_fairness = analysis['best_fairness_model'].get_fairness(selected_dataset.dataset_orig_test)\n",
    "\n",
    "print(\"[Best fairness] Accuracy on test set: \" + str(test_accuracy))\n",
    "print(\"[Best fairness] Fairness (True Positive Rate Difference) on test set: \" + str(test_fairness['eq_opp_diff']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 576x396 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "analyser.plot_results()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Task 3: Accuracy + Fairness Based Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Machine-Learning based model using accuracy+fairness criterion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy: 3\n",
      "Best fairness: 4\n",
      "Best accuracy+fairness: 0\n"
     ]
    }
   ],
   "source": [
    "analyser = Analysis(dataset=selected_dataset, C_values=C_values, solver=\"liblinear\", params={\"penalty\": \"l2\"}, mitigated=False, include_delta=True)\n",
    "analysis = analyser.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "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>model</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>accuracy_std</th>\n",
       "      <th>fairness</th>\n",
       "      <th>accuracy+fairness</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0001</td>\n",
       "      <td>0.655714</td>\n",
       "      <td>0.023647</td>\n",
       "      <td>-0.157834</td>\n",
       "      <td>0.826405</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0010</td>\n",
       "      <td>0.660000</td>\n",
       "      <td>0.024578</td>\n",
       "      <td>-0.238396</td>\n",
       "      <td>0.904111</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0100</td>\n",
       "      <td>0.687143</td>\n",
       "      <td>0.029067</td>\n",
       "      <td>-0.294951</td>\n",
       "      <td>0.989237</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.1000</td>\n",
       "      <td>0.695714</td>\n",
       "      <td>0.029485</td>\n",
       "      <td>-0.139268</td>\n",
       "      <td>0.846411</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.0000</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>-0.133386</td>\n",
       "      <td>0.840529</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    model  accuracy  accuracy_std  fairness  accuracy+fairness\n",
       "0  0.0001  0.655714      0.023647 -0.157834           0.826405\n",
       "1  0.0010  0.660000      0.024578 -0.238396           0.904111\n",
       "2  0.0100  0.687143      0.029067 -0.294951           0.989237\n",
       "3  0.1000  0.695714      0.029485 -0.139268           0.846411\n",
       "4  1.0000  0.692857      0.031623 -0.133386           0.840529"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "analysis['data_frame']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 576x396 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "analyser.plot_results(type=\"accuracy+fairness\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Accuracy and Fairness on test data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__[Model 5]: Best Accuracy__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test set: 0.64\n",
      "Fairness (True Positive Rate Difference) on test set: -0.0970833333333333\n"
     ]
    }
   ],
   "source": [
    "# Best accuracy+fairness\n",
    "test_accuracy = analysis['best_accuracy_fairness_model'].get_accuracy(X_test, y_test)\n",
    "test_fairness = analysis['best_accuracy_fairness_model'].get_fairness(selected_dataset.dataset_orig_test)\n",
    "\n",
    "print(\"Accuracy on test set: \" + str(test_accuracy))\n",
    "print(\"Fairness (True Positive Rate Difference) on test set: \" + str(test_fairness['eq_opp_diff']))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Fairness based model using accuracy+fairness criterion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy: 3\n",
      "Best fairness: 3\n",
      "Best accuracy+fairness: 4\n"
     ]
    }
   ],
   "source": [
    "analyser = Analysis(dataset=selected_dataset, C_values=C_values, solver=\"liblinear\", params={\"penalty\": \"l2\"}, mitigated=True, include_delta=True)\n",
    "analysis = analyser.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "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>model</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>accuracy_std</th>\n",
       "      <th>fairness</th>\n",
       "      <th>accuracy+fairness</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0001</td>\n",
       "      <td>0.655714</td>\n",
       "      <td>0.023647</td>\n",
       "      <td>-0.139876</td>\n",
       "      <td>0.809876</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0010</td>\n",
       "      <td>0.660000</td>\n",
       "      <td>0.024578</td>\n",
       "      <td>-0.103993</td>\n",
       "      <td>0.773993</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0100</td>\n",
       "      <td>0.687143</td>\n",
       "      <td>0.029067</td>\n",
       "      <td>-0.044911</td>\n",
       "      <td>0.752053</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.1000</td>\n",
       "      <td>0.695714</td>\n",
       "      <td>0.029485</td>\n",
       "      <td>0.003592</td>\n",
       "      <td>0.684979</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.0000</td>\n",
       "      <td>0.692857</td>\n",
       "      <td>0.031623</td>\n",
       "      <td>0.047363</td>\n",
       "      <td>0.641208</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    model  accuracy  accuracy_std  fairness  accuracy+fairness\n",
       "0  0.0001  0.655714      0.023647 -0.139876           0.809876\n",
       "1  0.0010  0.660000      0.024578 -0.103993           0.773993\n",
       "2  0.0100  0.687143      0.029067 -0.044911           0.752053\n",
       "3  0.1000  0.695714      0.029485  0.003592           0.684979\n",
       "4  1.0000  0.692857      0.031623  0.047363           0.641208"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "analysis['data_frame']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjgAAAGECAYAAAA7lVplAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAABV6klEQVR4nO3deXhMB//+8fckkViSEISqFkWp0pbUUjRirV0iqS3EUkttbam1qF1spWotqrYqIYkktqrYoraWUqWWp1pa1aKWkoRsM78//Jrv40ETmslJTu7XdbmaOTNzzj0+Rm/nnJljsdlsNkRERERMxMHoACIiIiIZTQVHRERETEcFR0RERExHBUdERERMRwVHRERETEcFR0REREzHyegAIpLxkpKSqFevHuXLl2fJkiVGx8kww4cPZ+/evRQsWPCe5YsWLaJo0aIPfZ6vry8rV67E3d3d3hHtLiUlhRUrVrBhwwZSUlJSZ/3OO+/g7OxsdDyRLEMFR8SEtm3bRvny5Tlx4gRnz56lTJkyRkfKMF27dqV79+6P9JzIyEg7pcl8Y8eO5a+//mL58uW4ubkRHx/P4MGDGTlyJNOnTzc6nkiWoYIjYkKrV6+mWbNmlCxZkuXLlzN+/HgAQkNDWbp0KQ4ODnh4eDB16lSKFSv2wOW//PILEyZMYOPGjQAcPHgw9facOXM4evQoly9fpnz58gwfPpzRo0dz9epVrly5QvHixZk1axaFChXi559/ZvTo0Vy7dg0HBwf69OlD0aJFeffdd9m5cycODg7cvn2b+vXrs3HjRgoVKgTc3VtTvXp1/P390/Waf/75Z8aPH098fDyXL1/mueeeY9asWbi4uFC+fHn279/Prl27CA0N5fbt27i6utK6dWu2bduGg4MD58+fJ1euXEydOpVy5cpx69YtJk2axJkzZ0hKSqJmzZoMHToUJycnZs+ezbZt28iVKxceHh5MnjyZIkWKPHT5327duoWPjw9bt27F09MTgLZt29KvXz/y5cvHlClTsFqtALz55ps0btz4ntf466+/smHDBr766itcXV0ByJs3L+PGjePIkSP/4k+MiPnoHBwRk/nxxx85evQoTZs2xc/Pj8jISK5fv86pU6f44IMP+OSTT9iwYQP169dnwYIFD12elt9++43169fzwQcfsGnTJipXrkxISAjbt28nd+7cqXtN3n33XZo0acKmTZtYtGgRM2fOpHz58hQoUIA9e/YAsGnTJmrWrJlabv7JsmXL8PX1Tf21bt06ANauXYufnx8hISF8+eWXXLhwgV27dj3w92flypWsXLkSgG+++Yb333+fjRs34uXllXpILzg4mIoVKxIeHk5ERATXr19n6dKl/P777yxfvpywsDDCw8OpXbs2x44de+jy/+bm5kajRo2IiooC4OzZs1y5cgVvb2/mzJlDt27dCA8PJzg4mAMHDtyX/YcffqBs2bKp5eZvnp6evPbaa2n+3onkJNqDI2Iyq1evpm7duhQoUIACBQrw1FNPERISgouLC6+++irFihUD7h7qAVi6dOkDlx88ePAft1O5cmWcnO7+FdKlSxcOHTrE0qVLOXfuHP/5z3946aWXuHHjBqdOnaJNmzYAFCtWjOjoaAA6duzI2rVr8fHxISQkhKFDhwLQpk0bEhMT+f333zlw4ADLly/Hy8uLMWPGpOZ70CGqIUOGsHfvXhYvXsy5c+e4fPky8fHx9z2ufPny9xSEihUr8sQTTwDw/PPPs23bNgB27drF999/T2hoKAB37twBoGjRojz33HO0bt2aOnXqUKdOHWrWrInVan3g8v/Vpk0bxo0bR/fu3QkLC8Pf3x8HBweaNm3K+PHj2bFjB7Vq1eLdd9+977kODg6pe3hE5J+p4IiYSHx8PBEREbi4uFC/fn0AYmNjWbVqFT169MBisaQ+9s6dO/z22284Ojo+cLnFYuG/L1WXlJR0z7by5s2b+vP06dM5duwYAQEB1KhRg+TkZGw2W2oB+u/1//TTTzz55JO0bNmSmTNncuDAAeLj46lWrRpA6h6ZRz1E9e6775KSkkLTpk2pW7cuv//+Ow+61N5/5wbInTt36s///ZqtVisfffRR6vlLN2/exGKx4ODgwGeffcb333/P/v37CQ4OpkaNGowaNeqhy/9b1apVSU5O5tixY2zcuJE1a9YA0L59e+rVq8fevXvZs2cPc+fOJSoqCjc3t9Tnvvjii/z000/ExsbeU9IuXbrE+++/z+zZs+95PSI5mQ5RiZjIhg0b8PDwYM+ePezYsYMdO3YQHR1NfHw8t27dYv/+/Vy+fBmANWvWMH36dGrUqPHA5QULFuTixYtcvXoVm82WuuflQb766iu6dOmCn58fhQoVYt++faSkpODq6krFihWJiIgA4Pfff6dDhw7cunWLPHny0KpVK0aMGEH79u3/9Wv/6quv6NevH82aNcNisfDdd9+RkpLy2Ot79dVXWbZsGTabjcTERPr06cNnn33GqVOnaNGiBWXKlOHNN9+ka9eunD59+qHLH6RNmzZMmDCB8uXL8+STTwJ3C87Jkyfx9/dnwoQJ3Lx5k7/++uue5xUtWpSWLVsyYsQIYmNjgbsFduzYsRQoUEDlRuS/aA+OiImsXr2abt264ejomLrM3d2doKAgdu7cyZAhQ+jRowdw97yN4OBgihYt+tDl7du3JyAgAE9PT+rWrfvQ7fbr149p06Yxf/58HB0d8fLy4pdffgFgxowZjBs3jpUrV2KxWJg0aVLqCbb+/v6p5878rylTpjzSax84cCD9+vUjf/785MmTh2rVqqVmeBwjR45k0qRJtGzZkqSkJGrVqkWPHj3IlSsXTZs2JSAggLx585I7d25GjRrFc88998DlD+Ln58fMmTOZOXNm6rLBgwcTHBzMrFmzcHBwoH///jz11FP3PXfMmDHMnz+f9u3b4+joSGJiIg0bNuStt9567NcqYkYW24P24YqI2JnNZmPx4sX89ttvjBs3zug4ImIy2oMjIoZo0KABBQsWTNcntkREHpX24IiIiIjp6CRjERERMR0VHBERETEdU56Dc+XKLbut28MjL9ev3//lYZI9aZ7moVmai+ZpHvaepaen2wOXaw/OI3Jyckz7QZJtaJ7moVmai+ZpHkbNUgVHRERETEcFR0RERExHBUdERERMRwVHRERETEcFR0RERExHBUdERERMRwVHRERETEcFR0REREzHlN9kLCIiYpRVq5azdu3nrF0bhYuLi9Fx0qVu3VeoVOnF1NulSpVm8ODhD3zs5s0bcHd359VXfTIr3mNRwREREclAX365hQYNXmP79i9p1qyl0XHSxd09P3PnLkrXY7PLa1LBERERUxk7dhQbNkRk6DpbtvRj7NiJaT7u228P8eSTT+HnF8D48aNp1qwlJ04cZ/bsGVitVjw9izBmzAR+/PHH+5YNGvQ2Q4aMoGTJUkREhHL16lWaNWvJsGEDcXfPT82atXn++UosXboYq9XK7du3GTNmIiVKlGTZsk/Ys2c3KSkp+PkFYLFYuHDhV/r1e4eUlBS6dQtk8eIVuLi48O23h4iMDGPcuMlpvp6PP57LqVM/cPPmX5QtW44RI8awZMlCChUqRIkSpViwYA65cuWiVavWrF69ksqVvTh79kcApkyZiaurKzNmzGD//oNYrVbatetI/foNCQ9fx5YtG3FwcKBChecZMGAIu3fv4LPPluPk5EThwp6MGxeMg8Pjn0mjgiMiIpJBNm6MpGVLP0qUKEWuXLk4ceI406cHM3bsJEqVeoaNGyM4d+7cA5c9zLVrV1my5DNy5cpFePg6Ro+eQOHCnqxY8Sk7d0ZTs2ZtDh7cx6JFy7BarXz88Vy6d+/FG290onfv/hw8uB8vr6pcuvQ706YFExsby59/XqF//17UqvUqgYGduXnzL/r375W6zf79B/L000/j5ubGrFnzsVqtBAW15cqVy/dkS0xMZPHi5QB88snHNGzYmIEDhzJu3CgOHNhLvnyuXLhwgQULlpCQkMCbb3ajWrUabN68gUGDhlGhQkXWrw8lOTmZbdu2EhgYRL16DdmyZSNxcXG4uT34QprpoYLzCM6d+5ndu89Qq1Z9cuXKZXQcERF5gLFjJ6Zrb0tGu3nzJvv37+X69WuEhoYQFxdLeHgI165dpVSpZwBo0cIP4IHL/pvN9n8/Fyv2ZOr/czw9PZk1azp58uTlypXLvPDCS/zyy3kqVKiIo6Mjjo6OvPXWQAAqV/bi66/3s3lzFF279qREiVLMnbvogXtwHnSIKjk5mevXrzNmzAjy5s3L7du3SU5OvucxJUqUvOd2uXLlAShSpCiJiYlcuvQjJ06cSC1PycnJ/PHHRUaMGM3q1Z/x++8fUbHiCwC89dZAVq5cRljYWkqWLEWdOnXT+1v/QPoU1SOYP382bdq0wcfnFb74YjO2//4TKCIiOdqXX26mRQtfPvxwHjNnzmHRouV8/fVBXFxc+PXXXwD47LNl7N69k8KFC9+3zNnZhatX/wTgzJlTqeu1WP7vf9VTp05ixIgxjBw5lsKFPQEoWbIUZ86cxmq1kpyczIABfUlMTKRly9Zs2BDJ9evXKVv22Ud+PQcO7OXy5UuMGxdMr179SEi4c9//9xwcLP/zrHtvlyxZiho1ajB37iJmz/6Y+vUbUrz4U0RFRTB48HvMnbuI//znNN9//x1RUevp3r0Xc+cuwmazEROz65Ez/zftwXkEw4ePInfuXCxevJjOndtTs2Ztxo6dSJUqLxsdTUREDLZhQyTvvz8+9Xbu3Lnx8alPwYIFmTx5PA4ODhQqVIi2bQMpUqTIfcucnXMxY8YUihZ9IrW8/K/GjZvSt29P8uTJjYdHIf788wrPPlueGjVq0qdPd6xWK61bv46zszMVK1bit99+pXXrNvesw8urKl5eVdN8PRUqVGTZsiX069cTi8XCk08W588/rzzS70nt2nU4dep7+vbtwe3b8dSpU4+8efNRpkxZ+vXrSd68efH09OT55ysRFxfH0KEDyJs3H3ny5KFWrVcfaVv/y2Iz4W6IK1du2W3dnp5u7N17iAkTRrN16xYA/P3bMGLE6Pt21UnW5+npZtc/L5J5NEtz0Tz/PavVSp8+3Zk5cw758rkalsPes/T0fPB5OjpE9RjKlSvPypUhrF+/iZdeqkJ4+Dpq1XqZcePe56+/bhgdT0REcriLF3/jjTc60aDBa4aWGyNpD84j+t8marVaWb8+lEmTxnHhwq94eHgwaNAwunbtgbOzs91ySMbQvxLNQ7M0F83TPLQHJ5tycHAgIKAt+/YdZvToCaSkWBk1ajivvlqNDRsidCKyiIiIAVRwMkju3Lnp3/8dDh48Ss+evblw4Ve6d+9M8+aN+Prrg0bHExERyVFUcDJYoUKFmDRpGl999TUtWvhy6NDXtGjRiDfeCOKnn84aHU9ERCRHUMGxk9Kly/LppyvZuHEbL79cjY0bI3n11WqMHDmUq1evGh1PRETE1FRw7Kx69Rps3hzNkiUrKF78KRYv/pgaNSozZ84s7ty5Y3Q8ERERU1LByQQWi4WWLf3Yu/cQEydOwdHRgQkTRlOr1suEhoZgtVqNjigiImIqKjiZyNnZmV69+nLw4FH69n2by5cv0bdvTxo3rsfevXuMjiciImIaKjgGKFDAg7FjJ7Jv32H8/V/nu++O0Lp1c4KC2nHmzGmj44mIiGR7KjgGKlGiJB9//Clbt+6kZs3abN26BR+fVxgyZCCXL19OewUiIiLyQCo4WUCVKi8TEbGZFSvW8MwzpVm+fAk1alRm5sxpxMfHGx1PREQk27FbwbFarYwePZp27doRFBTE+fPn77n/008/xd/fn4CAALZt2wbAnTt3eOuttwgMDKRnz55cu3YNgB07dhAQEEC7du1Yu3atvSIbymKx0KRJM3bvPsDUqTPJkyc3U6ZM5JVXqrB69WekpKQYHVFERCTbsFvBiY6OJjExkZCQEAYNGsSUKVNS77t58yYrVqxgzZo1fPrppwQHBwOwevVqypUrx+eff46fnx/z588nKSmJyZMn8+mnn7Jy5UpCQkL4888/7RXbcLly5aJbtx4cPHiUgQMHc+PGdd55py8NGnizc+d2o+OJiIhkC3YrOIcPH8bb2xuAypUrc/z48dT78uTJw5NPPsnt27e5ffs2FovlvufUqVOH/fv3c/bsWUqUKEH+/Plxdnbm5Zdf5ptvvrFX7CzDzc2d994bzYEDR2jXLpCTJ0/Qrl1r2rVrzYkTx9NegYiISA7mZK8Vx8bG4ur6f5dod3R0JDk5GSenu5ssVqwYzZs3JyUlhTfffDP1OW5ud68Kmi9fPm7dunXPsr+Xx8bG/uO2PTzy4uTkmNEvKdXDrlxqn209x5o1qxg+fAhDhgwhOjqaXbt20K1bN8aPH0/x4sUzLYtZZeY8xb40S3PRPM3DiFnareC4uroSFxeXettqtaaWm5iYGC5fvsz27XcPuXTv3h0vL697nhMXF4e7u/t964mLi7un8DzI9ev2OzHX3pd9f5jixcuwalUYO3dGM27c+3z66aesXr2aPn3eon//d3B11V8Ej8OoeUrG0yzNRfM0D3vP8mHlyW6HqLy8vIiJiQHg6NGjlCtXLvW+/Pnzkzt3bpydnXFxccHNzY2bN2/i5eXF7t27gbsl6OWXX6ZMmTKcP3+eGzdukJiYyKFDh6hSpYq9YmdpFouF+vUbsWPHXj78cC7u7vmZOXMaNWpUYfnyT0lOTjY6ooiISJZgsdlsNnus2Gq1MnbsWM6cOYPNZiM4OJiYmBhKlChBgwYNmD17Nnv27MHBwQEvLy+GDh3KnTt3GDZsGFeuXCFXrlzMmDEDT09PduzYwbx587DZbAQEBNCxY8d/3La9m2JW+VdFXFwcCxbMYe7cj4iPj6NcufKMHj2eRo2apJ7XJP8sK81T/h3N0lw0T/Mwag+O3QqOkXJKwfnbpUt/MG3aZFatWo7VauXVV+swduxEXnyxstHRsrysOE95PJqluWie5mG6Q1SSeYoWfYIZMz5i1679NGz4Gl99FUPDhnXo27cnFy78anQ8ERGRTKeCYyLPPVeBzz8PJTQ0ikqVXiQ0NISaNb2YMGEMN2/+ZXQ8ERGRTKOCY0J16tQlOjqGuXMXUqhQYebM+ZDq1V/ik08+JjEx0eh4IiIidqeCY1IODg60bduB/fu/ZdSosSQmJjFixFC8vauzcWMUJjz1SkREJJUKjsnlyZOHt99+l6+//o7u3Xvx66+/8MYbnWjZsjGHDn1tdDwRERG7UMHJIQoXLszkyR+wZ89BmjVryddfH6BZs4b07NmVc+d+NjqeiIhIhlLByWHKlHmWZctWERX1BV5eLxMZGU7t2lV5//33uH79mtHxREREMoQKTg71yiu12LJlB4sWLaVYsSdZuHAe1atXZv78OSQkJBgdT0RE5F9RwcnBLBYLfn4B7N17iHHjggEYO3YktWtXZf36UJ2ILCIi2ZYKjuDi4kKfPv35+uuj9O7dnz/++J0333yDJk3qsX//XqPjiYiIPDIVHEnl4VGQ8eOD+eqrb/Dz8+fIkW/x9W1Kly6B/Pjjf4yOJyIikm4qOHKfUqWeYdGiZWzZsp0aNWqyZctGvL2rM3z4IP7880+j44mIiKRJBUce6uWXqxEV9QXLln1OyZKl+PTTxVSv/hIffTSD27dvGx1PRETkoVRw5B9ZLBaaNWvBnj1fM3nydFxcnJk0aRw1a3oREvI5VqvV6IgiIiL3UcGRdMmVKxfdu7/JwYNHefvtd7l27SpvvdWbhg3rEBOzy+h4IiIi91DBkUfi7p6fUaPGsm/fYdq0ac/x48d4/fVWdOgQwMmTPxgdT0REBFDBkcf01FNPM2/eIqKjY/D29mH79m3Uq1eLQYPe5tKlP4yOJyIiOZwKjvwrL75YmdDQKD7/fB3PPluOlSuXUaNGFaZPn0xsbKzR8UREJIdSwZF/zWKx0LBhY3bu3MeMGbPJly8f06dP5pVXqvDZZ8tJSUkxOqKIiOQwKjiSYZycnAgK6srBg0cZNGgYsbG3ePfdt6hXrxbbt3+pSz+IiEimUcGRDOfq6sqwYSM5cOAIHTt25syZ03To8Dqvv+7L998fMzqeiIjkACo4YjdPPFGMDz+cy44de6lfvyF79uyiYUNv+vd/k99+u2B0PBERMTEVHLG755+vyJo14axdG8Hzz1di7drV1KzpRXDweG7duml0PBERMSEVHMk0devWJzo6htmzF+DhUZBZsz6gRo3KfPrpYpKSkoyOJyIiJqKCI5nK0dGR9u07sn//t4wYMZrbt+8wfPgg6tSpwZYtm3QisoiIZAgVHDFE3rx5GTBgMAcPHqVr1+6cO/czXbp0wNe3Kd9+e8joeCIiks2p4IihihQpwrRpHxITc5AmTZpx4MA+mjSpz5tvduP8+XNGxxMRkWxKBUeyhGefLceKFWuIiNhM5cpVWL8+jNq1qzJmzEhu3LhudDwREclmVHAkS6lV61W++GInH3+8hKJFn2DBgjlUr/4SCxfOIzEx0eh4IiKSTajgSJbj4OCAv38b9u49xJgxE7Fabbz//nvUrl2VqKj1OhFZRETSpIIjWVbu3Lnp1+9tvv76KG++2ZeLF3+jR48uNGvWkIMHDxgdT0REsjAVHMnyChYsxIQJU9iz52tatvTj8OFvaNnyNbp168RPP/1odDwREcmCVHAk2yhdugxLlqxg48ZtVK1anU2bonj11eqMGDGEq1evGh1PRESyEBUcyXaqV6/Bpk3bWLJkJU899TSffLKQ6tVfYs6cWdy5c8foeCIikgWo4Ei2ZLFYaNnSl6+++oZJk6bi5OTIhAmjqVXrZUJDQ7BarUZHFBERA6ngSLbm7OxMz559+Prr7+jffwBXrlymb9+evPZaXb76KsboeCIiYhAVHDGF/PkLMHr0ePbtO0xAQFuOHTuKv38LOnVqy5kzp42OJyIimcxis9OXilitVsaOHcvp06dxdnZm4sSJlCxZEoCTJ08SHByc+tijR48yb9489uzZw6lTpwC4cuUK7u7urF27lokTJ/Ltt9+SL18+AObPn4+bm9tDt33lyi17vCQAPD3d7Lp+yRhHj37L2LGj2LfvKxwdHenYsQtDh46gSJEi9zxO8zQPzdJcNE/zsPcsPT0f3Aec7LXB6OhoEhMTCQkJ4ejRo0yZMoUFCxYAUKFCBVauXAnAli1bKFKkCHXq1KFOnToAJCUlERgYyIQJEwA4ceIEn3zyCQULFrRXXDGZypW9WL9+E19++QXjx7/PihWfEha2lv7936F37/6pZVlERMzJboeoDh8+jLe3NwCVK1fm+PHj9z0mPj6eOXPmMHLkyHuWf/bZZ9SuXZvy5ctjtVo5f/48o0ePpn379oSGhtorspiMxWKhceOm7N59gGnTPiRPnjxMnTqJmjW9WL36M1JSUoyOKCIidmK3PTixsbG4urqm3nZ0dCQ5ORknp//bZGhoKE2aNLlnz0xiYiJr1qxJLTLx8fF06tSJbt26kZKSQufOnalUqRLPPffcQ7ft4ZEXJydHO7yqux62O0yyriFDBtC7d3emTZvGjBkzeOedvnzyyQKGDh1Kq1atcHd3NzqiZAC9N81F8zQPI2Zpt4Lj6upKXFxc6m2r1XpPuQHYsGEDs2fPvmfZ/v37qVatWuo5Nnny5KFz587kyZMHgFdeeYVTp079Y8G5fj0+o17GfXRcOHt7++2hvP56R6ZOncSaNasICgrCxcWFBg1ew9e3NY0aNbmnmEv2ofemuWie5mHUOTh2O0Tl5eVFTMzdj+kePXqUcuXK3XP/rVu3SExMpFixYvcs37dvX+q5OADnzp2jQ4cOpKSkkJSUxLfffkvFihXtFVtygCefLM5HH81n//7DjBs3jlKlnmHz5g28+eYbVKxYhu7dO7NhQwTx8fYryiIiYl9224PTqFEj9u7dS/v27bHZbAQHB7N06VJKlChBgwYN+PnnnylevPh9z/v555/x8/NLvV2mTBl8fX1p27YtuXLlwtfXl2effdZesSUHKV26LKNHj6ZPn4GcOnWSyMhwIiPD2bAhgg0bIsibNx+NGzehVSt/GjRoRO7cuY2OLCIi6WS3j4kbSR8Tl/T633nabDZ++OEEkZHhRESEce7czwC4urrRuHFT/PwCqFu3Pi4uLkZFlofQe9NcNE/zMOoQlQrOI9Kbzlz+aZ42m43vv/+OiIhwoqLW88sv5wFwd89P06bN8fPzx9u7Ls7OzpkZWR5C701z0TzNQwUnA6ngSHqld542m40jRw4TGbmeqKj1/PbbBQAKFChA8+ataNWqNd7ePvedSC+ZR+9Nc9E8zUMFJwOp4Eh6Pc48rVYrhw9/Q2RkOFFREfzxx+8AFCpUiGbNWuHn50+tWq/i6Gi/ryqQ++m9aS6ap3mo4GQgFRxJr387T6vVytdfHyAiIowNGyK5cuUyAIULe9KypS++vv7UqFFTZScT6L1pLpqneajgZCAVHEmvjJxnSkoK+/fvJTJyPZs2RfLnn38CULToE/+/7ARQrVp1HBx0jVt70HvTXDRP81DByUAqOJJe9ppncnIye/fuITIynE2borh+/Tpw9zt4Wrb0w8/PHy+vqlgslgzfdk6l96a5aJ7moYKTgVRwJL0yY55JSUns2bOLyMj1bN68kb/+ugHA00+XoFWr1vj6tuall6qo7PxLem+ai+ZpHio4GUgFR9Irs+eZmJjIrl3biYxcz5Ytm4iNvbvtkiVL4evrj6+vP5UqvaCy8xj03jQXzdM8VHAykAqOpJeR87xz5w47d24nMjKML77YQnz83Wu3lSlTFl/f1vj6BvDccxVUdtJJ701z0TzNQwUnA6ngSHpllXnevn2b6OgviYpaz7ZtX6ReB6tcufKpe3bKlStvcMqsLavMUjKG5mkeKjgZSAVH0isrzjMuLo7o6K1ERISzffuX3LlzB4AKFSri69saPz9/Spcua3DKrCcrzlIen+ZpHio4GUgFR9Irq88zNvYWW7duITJyPTt2bCMxMRGASpVexM/Pn1atWlOq1DMGp8wasvos5dFonuahgpOBVHAkvbLTPG/e/IsvvthMZGQ4u3btICkpCYDKlavQqpU/vr6tefrpEganNE52mqWkTfM0DxWcDKSCI+mVXed548Z1tmzZREREGDExu0hJSQHg5Zer4ut7d8/Ok08WNzhl5squs5QH0zzNQwUnA6ngSHqZYZ5Xr15l8+YNREau56uvdmO1WgGoXv0V/Pz8adnSj6JFnzA4pf2ZYZbyfzRP81DByUAqOJJeZpvnlStX2LQpisjIcPbt+wqbzYbFYqFmzdq0atWaFi18KVKkiNEx7cJss8zpNE/zUMHJQCo4kl5mnuelS5fYuDGCiIhwDh7cD4CDgwO1a3vj6+tP8+atKFSokMEpM46ZZ5kTaZ7moYKTgVRwJL1yyjx///0iUVHriYxcz6FDXwPg6OiIt7cPvr7+NGvWAg+Pggan/HdyyixzCs3TPFRwMpAKjqRXTpznhQu/EhUVQWRkGEeOfAuAk5MTdevWp1Wr1jRt2pz8+QsYG/Ix5MRZmpnmaR4qOBlIBUfSK6fP8/z5c0RGricqaj3Hjh0FwNnZmXr1GuDr60/jxk1xc3M3NmQ65fRZmo3maR4qOBlIBUfSS/P8Pz/99CORkXcPY/3ww3EAXFxcaNDgNXx9W9OoURNcXV0NTvlwmqW5aJ7moYKTgVRwJL00zwf7z3/OEBkZTmRkOKdPnwIgT548NGzYGD8/fxo0eI28efManPJemqW5aJ7moYKTgVRwJL00z7SdOnWSiIgwIiPDOXv2RwDy5s1H48ZNaNXKnwYNGpE7d26DU2qWZqN5mocKTgZSwZH00jzTz2azceLEcaKi1hMREca5cz8D4OrqRpMmzfD19adu3fq4uLgYkk+zNBfN0zxUcDKQCo6kl+b5eGw2G99//x0REeFERa3nl1/OA+Dunp+mTZvj5+ePt3ddnJ2dMy2TZmkumqd5qOBkIBUcSS/N89+z2WwcOXI49dNYv/12AYACBQrQvHkrWrVqjbe3D05OTnbNoVmai+ZpHio4GUgFR9JL88xYVquVQ4e+ISoqnKioCP7443cAChUqRLNmrfDz86dWrVdxdHTM8G1rluaieZqHCk4GUsGR9NI87cdqtfL11weIiAhjw4ZIrly5DEDhwp60bOmLr68/NWrUzLCyo1mai+ZpHio4GUgFR9JL88wcKSkp7N+/l8jI9WzcGMHVq1cBKFr0if9fdgKoVq06Dg4Oj70NzdJcNE/zUMHJQCo4kl6aZ+ZLTk5m7949REaGs2lTFNevXwfgySeL07KlH35+/nh5VcVisTzSejVLc9E8zUMFJwOp4Eh6aZ7GSkpKYs+eXURGrmfz5o389dcNAJ5+ugStWrXG17c1L71UJV1lR7M0F83TPFRwMpAKjqSX5pl1JCYmsmvXdiIj17NlyyZiY+/OpWTJUvj6+uPr60+lSi88tOxoluaieZqHCk4GUsGR9NI8s6Y7d+6wc+d2IiPD+OKLLcTHxwFQpkxZfH1b4+sbwHPPVbin7GiW5qJ5mocKTgZSwZH00jyzvtu3bxMd/SWRkeFs2/YFt2/fBqBcufKpe3bKlSuvWZqM5mkeKjgZSAVH0kvzzF7i4uLYtu0LIiPXs337l9y5cweAChUq0rVrZzp16kGuXLkMTikZQe9N81DByUAqOJJemmf2FRt7i61btxAZGc6OHdEkJibSunUA8+d/YpcvEpTMpfemeRhVcOz23elWq5WxY8dy+vRpnJ2dmThxIiVLlgTg5MmTBAcHpz726NGjzJs3jxdffJHGjRtTrlw5ABo2bEiXLl1Yu3Yta9aswcnJiT59+lCvXj17xRaRbMLV1Y2AgLYEBLTlxo3rdOsWyPr1Ybi45GbWrHn/6jt1RCT7S1fBSUxMxNnZmfPnz/Pzzz9Tp06dNP/yiI6++y+qkJAQjh49ypQpU1iwYAEAFSpUYOXKlQBs2bKFIkWKUKdOHfbt20eLFi14//33U9dz5coVVq5cSVhYGAkJCQQGBlK7du1MvYifiGRtBQp4sHnzZurVq8+aNatwdnZh+vQPH/m7dETEPNL8J87cuXMZNWoUFy9epGPHjixbtozRo0enueLDhw/j7e0NQOXKlTl+/Ph9j4mPj2fOnDmMHDkSgOPHj3PixAk6derE22+/zeXLlzl27BhVqlTB2dkZNzc3SpQowalTpx71dYqIybm7u7NmTTiVKr3IihWf8v77wzHhEXgRSac09+Ds2LGDNWvWsGzZMlq1asXQoUPx9/dPc8WxsbG4urqm3nZ0dCQ5OfmeKwqHhobSpEkTChYsCEDp0qWpVKkStWrVIioqiokTJ9KgQQPc3P7v+Fq+fPmIjY39x217eOTFycl+x+AfdrxPsifN0zyefbYEO3dup27duixatAAPD3cmT56sPTnZlN6b5mHELNMsOFarFWdnZ3bu3MmAAQOwWq2pH9P8J66ursTFxd2znv8uNwAbNmxg9uzZqbdfeeUV8uTJA0CjRo2YPXs2vr6+96wnLi7unsLzINevx6eZ73HpxDdz0TzN4/9m6UJISCS+vk2YOnUqVqsDQ4a8Z3Q8eUR6b5qHUScZp3mIqmbNmrRo0YKkpCSqVatGp06dqF+/fpob9PLyIiYmBrh7EvHfJw7/7datWyQmJlKsWLHUZaNGjWLr1q0A7N+/n4oVK/Liiy9y+PBhEhISuHXrFmfPnr1vXSIi/61IkSKEh2+kZMlSTJ8+mdmzZxodSUQyWbo+Jn7x4kWKFi2Ko6Mjp06d4rnnnktzxX9/iurMmTPYbDaCg4OJiYmhRIkSNGjQgGPHjvHxxx8zf/781Of8+uuvjBgxAoA8efIwceJEihQpwtq1awkJCcFms/Hmm2/SuHHjf9y2PiYu6aV5mseDZvnrr7/g69uUCxd+ZcKEybz5Zj+D0smj0nvTPLLs9+AcO3aMw4cP07FjR3r37s0PP/zAuHHj0iwZRlLBkfTSPM3jYbP86aez+Po25dKlP5g27UO6du1uQDp5VHpvmkeWPUQ1ceJEKlWqxNatW8mdOzfr169n0aJFGR5QRMQeSpcuQ3j4RgoX9mTo0IGsXv2Z0ZFEJBOkWXCsVivVqlVj165dvPbaaxQrVoyUlJTMyCYikiGefbYcoaFReHh4MGBAP8LD1xkdSUTsLM2CkydPHj799FMOHDhAvXr1WL58Ofny5cuMbCIiGeb55yuybl0kbm7u9OvXi40bo4yOJCJ2lGbB+eCDD4iPj2fu3Lnkz5+fy5cvM2PGjMzIJiKSoV58sTJr1oSRO3ce3nyzG19+ucXoSCJiJ2kWnKJFi/LKK69w6tQpEhMTqVu3Lk888URmZBMRyXBVq1Zn9epQcuXKxRtvBLFz53ajI4mIHaRZcJYvX85HH33EsmXLiIuLY/To0SxZsiQzsomI2MUrr9RixYo1WCwWunYNZN++r4yOJCIZLM2Cs379epYsWUKePHnw8PAgNDSUsLCwzMgmImI3derUZdmyVSQnJxMY2Iavvz5odCQRyUBpFhwHB4d7rtzt4uKCo6P9rvMkIpJZGjR4jcWLl5OQcIcOHQI4evRboyOJSAZJs+BUr16dqVOncvv2baKjo+nTpw+vvPJKZmQTEbG7Zs1asGDBJ8TFxdK2rR/Hj39vdCQRyQBpFpyhQ4dSsmRJypcvT0REBD4+PgwbNiwzsomIZAo/vwA++mg+f/31F23b+nL69CmjI4nIv5Tm1cQdHBxo0aIFPj4+/H1Vh8uXL/Pkk0/aPZyISGZp1y6QhIQEBg9+h4CAlkRFbaF06bJGxxKRx5Rmwfn4449ZtGgRBQoUwGKxYLPZsFgsbN+uj1aKiLl07tyNxMQERowYir9/SyIjt1CyZCmjY4nIY0iz4ISGhhIdHU3BggUzI4+IiKF69OjNnTsJjB//PgEBd0tO8eJPGR1LRB5RmufgFCtWjPz582dGFhGRLKF//3cYNmwkv/xynoCAlly69IfRkUTkEaW5B6dUqVIEBgZSo0aNez4u3r9/f7sGExEx0rvvDiUhIYFZsz4gIKAl69dvxtPT0+hYIpJO6bpUg7e39z3lRkTE7CwWC++99z69e/fnzJnTtGnjy/Xr14yOJSLplOYenN9++43JkydnRhYRkSzFYrEwbtwkEhLusHTpJ7Rr15rQ0Cjc3XXYXiSrS3MPzpkzZ4iLi8uMLCIiWY7FYmHy5A8IDAzi6NEjtG8fQGzsLaNjiUga0vU9OPXq1eOZZ57BxcUldfmKFSvsGkxEJKtwcHBgxozZJCQkEBa2lk6d2vH556HkzZvX6Ggi8hBpFpwhQ4ZkRg4RkSzN0dGROXM+JjExkQ0bIujaNZAVK9aQO3duo6OJyAM89BDViRMngLu7Zx/0S0Qkp3FycmLBgk9o3Lgpu3btoEePziQmJhodS0Qe4KF7cNasWcOECROYPXv2ffdZLBYdohKRHMnZ2ZlPPllB587t+fLLL3jzzTdYvHgZTk5p7hAXkUxksf19gSkTuXLFficAenq62XX9krk0T/PI7FnGx8fTsWMb9u7dg7//68ybtxhHR8dM277Z6b1pHvaepaen2wOXp/lPjkOHDrFkyRLi4+Ox2WxYrVYuXrzIjh07MjykiEh2kTdvXlauDKFdu9aEh4fi4pKbDz+ci4NDmh9OFZFMkOY7cdSoUTRs2JCUlBQ6duxIyZIladiwYWZkExHJ0lxdXVm9OpQqVbxYvfozhg0bhAl3iotkS2kWnNy5cxMQEED16tVxd3dn4sSJfPPNN5mRTUQky3N3z09IyHoqVXqR5cuXMHr0eyo5IllAmgXHxcWFGzdu8Mwzz/Ddd99hsViIj4/PjGwiItlCgQIerF0bQfnyz7Fw4XwmTRqnkiNisDQLTteuXRk4cCD16tUjIiKC5s2bU6lSpczIJiKSbRQuXJjQ0A2UKVOW2bNnMmPGVKMjieRo6foUlc1mS91zc+7cOZ577rksfSKdPkUl6aV5mkdWmeXFi7/RqlVTfvnlHKNGjePttwcaHSlbyirzlH/PqE9RPbSl9OjRI/Xnv8+5yZs3L88//3yWLjciIkZ68snihIdvoHjxp5g4cQyLFs03OpJIjvTQpvLnn3+m/qyriYuIpF+JEiUJC4uiaNEnGDVqOMuXf2p0JJEcJ127YnSynIjIoylduixhYRsoXLgwQ4YMYM2aVUZHEslRHlpw/vt6U7r2lIjIoytXrjzr1kXh4eHBgAH9WL8+1OhIIjnGQ7/J+OLFi7z33nv3/fw3HbYSEUlbxYqVWLs2An//lvTt2xNnZxeaN29pdCwR03towRk+fHjqz9WrV8+UMCIiZvTSS1VYsyaMtm1b06tXV5YtW0WjRk2MjiViaum+2OalS5coWrSovfNkCH1MXNJL8zSP7DDL/fv30r69P1arlZUrQ6hbt77RkbKs7DBPSZ8s9zHx/9W7d+8MCyMikhPVrFmbFSvWANClSwf27fvK4EQi5mW3L7SxWq2MHj2adu3aERQUxPnz51PvO3nyJEFBQam/XnjhBWJiYrh48SJdu3YlKCiITp068dNPPwGwbNkymjdvnvr4v5eLiGQ3Pj71+PTTlSQnJ9OxY1u++eag0ZFETOmh5+D8rX79+lgsFi5fvpz68/bt29NccXR0NImJiYSEhHD06FGmTJnCggULAKhQoQIrV64EYMuWLRQpUoQ6deowbNgwOnXqRMOGDdmzZw8zZ85k7ty5HD9+nKlTp+oSESJiCo0aNWHRomX06NGZ9u0DCAuLonJlL6NjiZhKmntwduzYwfbt2ylTpkzqz+lx+PBhvL29AahcuTLHjx+/7zHx8fHMmTOHkSNHAjBs2DB8fHwASElJwcXFBYATJ06waNEiOnTowMKFC9P3ykREsrDmzVsyf/5i4uJiadvWjxMn7v87UkQeX5p7cB5XbGwsrq6uqbcdHR1JTk7Gyen/NhkaGkqTJk0oWLAgQOp/f/rpJ6ZOncq8efMAaN68OYGBgbi6utK/f3927txJvXr1HrptD4+8ODk52uNlAQ8/oUmyJ83TPLLbLHv16oaLiwNdu3albVtfdu/eTYUKFYyOlWVkt3nKwxkxyzQLzpUrV/D09CQ8PPyRVuzq6kpcXFzqbavVek+5AdiwYQOzZ8++Z9mBAwcYN24c06ZNo3Tp0thsNrp06YKb293fHB8fH3744Yd/LDjXr8c/UtZHoTP7zUXzNI/sOstmzfz54IO/GDz4HerVq09k5GZKly5rdCzDZdd5yv2y7KeoOnXqRK9evdi6dStJSUnp3qCXlxcxMTEAHD16lHLlyt1z/61bt0hMTKRYsWKpyw4cOMCkSZP45JNPeOGFF4C7e4JatGhBXFwcNpuNgwcP6lwcETGVzp27MWnSVC5d+oOAgFb88sv5tJ8kIv8ozT04W7du5dChQ6xfv54PPvgAHx8fWrdunVpAHqZRo0bs3buX9u3bY7PZCA4OZunSpZQoUYIGDRrw888/U7x48XueExwcTFJSUuqXDD7zzDOMHz+egQMH0rlzZ5ydnalZs2bqeToiImbRs2cf7txJYMKE0fj7tyQqagtPPlk87SeKyAOl+4v+EhIS2LJlCx9++CEWi4WCBQsyevRoKleubOeIj05f9CfppXmah1lm+cEHU5g2LZjSpcsQGbmFokWfMDqSIcwyTzHuEFWae3D27dtHZGQk+/btw8fHhw8//BAvLy9Onz5Nz549Uw9DiYjIvzdo0DASEhL46KMZvP56K9av30zhwoWNjiWS7aRZcObNm8frr7/O2LFjyZMnT+ry8uXL88Ybb9g1nIhITmOxWBgxYjR37txm4cL5tGnjS3j4Bjw8ChodTSRbSfMk44ULFxIfH0+ePHm4dOkSH330Ebdv3waga9eu9s4nIpLjWCwWxo+fTNeu3Tlx4nvat/fn5s2/jI4lkq2kWXAGDx7M5cuXAciXLx9Wq5WhQ4faPZiISE5msViYMmUGHTp04siRb+nQ4XViY2ONjiWSbaRZcC5evMjAgQOBu99tM3DgQH755Re7BxMRyekcHByYOXMO/v5t+OabgwQFtSM+3n7f8yViJmkWHIvFwunTp1Nvnz179r4v7BMREftwdHRk7tyFtGjhy969e+jaNZA7d+4YHUsky0uzqQwbNow33niDokWLAnD9+nWmTZtm92AiInKXk5MTH3+8hDfeSODLL7+gZ88uLFmyEmdnZ6OjiWRZ6foenMTERM6cOYOTkxOlS5fO8m8qfQ+OpJfmaR45YZZ37twhKKgdu3fvpEULXxYtWmraPeo5YZ45RZb9HpyffvqJzz//nPj4eGw2G1arlQsXLrBq1aoMDykiIg+XO3duli9fTWDg62zcGEn//m8yb94iHB3td3FhkewqzXNwBg4ciLu7OydPnqRChQpcvXqVZ599NjOyiYjI/8ibNy+ffbaWatVqEB6+jkGD3sZqtRodSyTLSbPgWK1W3n77bby9vXn++eeZP38+x44dy4xsIiLyAK6urqxeHUrlylX4/POVDB8+iHRedUckx0iz4OTJk4fExERKlSrFiRMncHZ2JiEhITOyiYjIQ7i75yckZD0VK77AsmVLGD16hEqOyH9Js+C0atWK3r17U7duXT777DN69OiR+okqERExjodHQdati6R8+edYuHAewcHjVXJE/r80TzKuWrUqfn5+uLq6snLlSr7//ntq166dGdlERCQNhQsXJjQ0Cl/fpnz00Qxy587NoEHDjI4lYrh0nWTs6uoKwBNPPEGjRo3Imzev3YOJiEj6FC36BGFhGyhRoiRTp05i7tyPjI4kYrg09+CULVuWuXPn8tJLL5E7d+7U5dWqVbNrMBERSb/ixZ8iLGwDvr5NGT/+fXLndqFHj95GxxIxTJoF58aNGxw8eJCDBw+mLrNYLKxYscKuwURE5NGULFmK8PANtGrVlBEjhuLs7ELnzt2MjiViiDQLzsqVKzMjh4iIZIDSpcsSFraB1q2bMWTIAJydnWnfvqPRsUQyXZoFJygoCIvFct9y7cEREcmaypd/jrVrI/H3b86AAf3InTs3fn4BRscSyVRpFpy33nor9efk5GS2b9+Ou7u7XUOJiMi/U6nSC6xdG0FAQCv69OlBrlzONG/e0uhYIpkmXRfb/F9t2rRh3bp19siTIXSxTUkvzdM8NMsH++abg7Rp40dSUiLLl39Ow4aNjY6ULpqneRh1sc00PyZ+8eLF1F+//fYbu3fv5saNGxmdT0RE7KBatRp8/vk6nJyc6NatE7t37zQ6kkimSPMQVadOnVJ/tlgsFCxYkFGjRtk1lIiIZJxatV5l+fLVBAW1o3Pn9qxZE07NmvrCVjG3dB2iSkpKIleuXCQlJZGUlJTlv+hPh6gkvTRP89As0/bll1vo1q0Tzs4urFsXQdWq1Y2O9FCap3lk2UNUW7Zswd/fH4Dff/+dpk2bEh0dnbHpRETE7l57rSkLFy7lzp3btG8fwHffHTE6kojdpFlw5s+fz9KlSwEoUaIE4eHhzJkzx+7BREQk47Vo0Yp58xZx69ZN2rb148SJ40ZHErGLNAtOUlIShQsXTr1dqFAhXa1WRCQb8/dvw0cfzef69eu0adOKM2dOGx1JJMOleZLxyy+/zLvvvkvLlne/P2Hz5s1UrlzZ3rlERMSO2rfvSEJCAkOGDCAgoCWRkVsoXbqM0bFEMkyae3DGjBlDxYoVCQkJISwsjEqVKulTVCIiJtClyxtMnDiFS5f+ICCgJb/8ct7oSCIZJl2HqHLnzs3HH3/M+++/z40bN0hJScmMbCIiYme9evVl1Khx/PbbBQICWnLx4m9GRxLJEGkWnEGDBnH58mUA8uXLh9VqZejQoXYPJiIimePttwcyZMh7nD9/joCAlly6dMnoSCL/Wrq+yXjgwIEAuLq6MnDgQH755Re7BxMRkcwzePBw3nprIGfP/kibNq24evWq0ZFE/pU0C47FYuH06f87w/7s2bM4OaV5brKIiGQjFouFUaPG0qtXH06dOkmbNr7cuHHd6Fgijy3NpjJs2DDeeOMNihYtCsD169eZPn263YOJiEjmslgsTJgwhYSERJYvX0K7dq0JDY3Czc3d6GgijyzNPTi1atVi586djB07lvr161OkSBF69uyZGdlERCSTWSwWpk6dQfv2HTly5Fs6dHid2NhYo2OJPLI09+D8+uuvhISEEB4ezs2bN+nduzcLFizIjGwiImIABwcHPvxwLgkJd1i/PozOnduzatU68uTJY3Q0kXR76B6cbdu20b17d9q0acNff/3F9OnTKVKkCP3796dgwYJprthqtTJ69GjatWtHUFAQ58//3/crnDx5kqCgoNRfL7zwAjExMVy7do033niDwMBABgwYwO3btwFYu3Yt/v7+tG3blp07d2bAyxYRkX/i6OjI3LmLaN68FV99FUPXroEkJCQYHUsk3R66B+ett96iSZMmhISEULJkSeDursv0io6OJjExkZCQEI4ePcqUKVNS9/xUqFCBlStXAncv5lmkSBHq1KnDxIkTadGiBf7+/ixatIiQkBCaN2/OypUrCQsLIyEhgcDAQGrXro2zs/O/ed0iIpKGXLlysXDhp3Tr1pFt27bSo0dnlixZqb9/JVt46B6cqKgoihUrRmBgIG3btmX58uWP9AV/hw8fxtvbG4DKlStz/Pj9F3SLj49nzpw5jBw58r7n1KlTh3379nHs2DGqVKmCs7Mzbm5ulChRglOnTj3SixQRkcfj7OzMkiUr8fGpx9atW+jTpwfJyclGxxJJ00P34JQrV45hw4YxePBgdu7cyfr16/nzzz/p1asXHTt2xMfH5x9XHBsbi6ura+ptR0dHkpOT7/mIeWhoKE2aNEk95BUbG4ubmxtw90sFb926dc+yv5endcKbh0denJwc//Ex/4anp1vaD5JsQ/M0D83SXtzYvHkjTZs2ZcOGCNzd87F8+XIcHe339yxonmZixCzTPMnY0dGRhg0b0rBhQ65du0ZkZCQzZsxIs+C4uroSFxeXettqtd73/TkbNmxg9uzZ9z0nd+7cxMXF4e7uft964uLi7ik8D3L9enxaL+uxeXq6ceXKLbutXzKX5mkemqX9LVu2mrZtW7Nq1SrAkRkzZuPgkOaHcR+L5mke9p7lw8rTI/3JLFiwIN26dSMqKirNx3p5eRETEwPA0aNHKVeu3D3337p1i8TERIoVK3bPc3bv3g1ATEwML7/8Mi+++CKHDx8mISGBW7ducfbs2fvWJSIi9ufq6saaNWG89FIVVq1awXvvDcZmsxkdS+SB7PaVxI0aNWLv3r20b98em81GcHAwS5cupUSJEjRo0ICff/6Z4sWL3/OcPn36MGzYMNauXYuHhwczZswgb968BAUFERgYiM1mY+DAgbi4uNgrtoiI/AN39/yEhITj79+SpUs/wcUlN+PGTXqkD6GIZAaLzYT12967wrTb1Dw0T/PQLDPXlStXaN26GWfOnGbgwMG8997oDF2/5mke2eIQlYiICICnpydhYRt45pnSfPjhB8ycOc3oSCL3UMEREZHHUrToE4SHb6REiZJMmTKRefNmp/0kkUyigiMiIo+tePGnCAvbwJNPFmfcuFEsWbLQ6EgigAqOiIj8SyVLliIsLIoiRYry3ntDWLlymdGRRFRwRETk3ytT5lnCwjZQqFAhBg9+h5CQz42OJDmcCo6IiGSI8uWfY926KPLnz8877/QlIiLM6EiSg6ngiIhIhqlU6QVCQtaTL58rffr0YPPmjUZHkhxKBUdERDJUlSovs3p1GC4uuenZswvR0VuNjiQ5kAqOiIhkuOrVa7Bq1VocHR3p1q0TMTG7jI4kOYwKjoiI2EXt2t4sX74am81G587tOXBgn9GRJAdRwREREbupV68BS5asJDExkQ4dXufQoa+NjiQ5hAqOiIjYVePGTVm4cCl37tymffsAjh07anQkyQFUcERExO5atvRl7tyF3Lp1k7Zt/fjhhxNGRxKTU8EREZFMERDQllmz5nHt2jVef70V//nPGaMjiYmp4IiISKbp0KETU6fO5M8/r+Dv34KffjprdCQxKRUcERHJVN269WD8+GAuXfqD119vxa+//mJ0JDEhFRwREcl0vXv3Z9SosVy48Cv+/i34/feLRkcSk1HBERERQ7z99rsMGjSM8+fP4e/fgkuXLhkdSUxEBUdERAwzdOgI+vcfwNmzP9KmTSuuXr1qdCQxCRUcERExjMVi4f33x9GzZ29OnTpJ27Z+3Lhx3ehYYgIqOCIiYiiLxcLEiVMJCurG999/R/v2/ty8edPoWJLNqeCIiIjhLBYL06d/SLt2gXz77WHq1KmjLwOUf0UFR0REsgQHBwdmzZpHly7d+e6772jUqA6zZn1AcnKy0dEkG1LBERGRLMPR0ZHp0z9k06ZNFCpUmODg8TRv3pAzZ04bHU2yGRUcERHJcpo1a0ZMzAFef70dR458S4MGrzJv3mxSUlKMjibZhAqOiIhkSQUKeDB//mKWLfscNzd3xo0bRatWTfjppx+NjibZgAqOiIhkac2atWDPnq/x9fXnm28OUq9ebRYvXoDVajU6mmRhKjgiIpLlFSpUiMWLl7F48TLy5MnDyJHD8PdvwblzPxsdTbIoFRwREck2fH39iYn5mqZNW7Bv31fUrVuLZcuWYLPZjI4mWYwKjoiIZCtFihRh2bJVzJ+/mFy5cjF06EDatPHjwoVfjY4mWYgKjoiIZDsWi4XXX2/Hnj0HadjwNWJiduLjU5PPP1+pvTkCqOCIiEg29sQTxVi1ah0ffTQfgAED+tGxYxv++ON3g5OJ0VRwREQkW7NYLHTo0Indu/fj41OP6Ogv8fauwbp1a7Q3JwdTwREREVN46qmnWbs2gunTZ5GUlES/fr3o0iWQy5cvGx1NDKCCIyIipmGxWOjS5Q12795P7drefPHFJurUqU5kZLjR0SSTqeCIiIjplCxZirCwDQQHT+P27dv07NmVHj26cPXqVaOjSSZRwREREVNycHCgR4/e7Ny5l2rVahAVtR5v7+ps3rzR6GiSCZzstWKr1crYsWM5ffo0zs7OTJw4kZIlS6bev3v3bubNm4fNZqNixYqMGTOGxYsXs2fPHgBu3rzJn3/+yd69e1m2bBnr1q2jYMGCAIwbN47SpUvbK7qIiJhI6dJliYr6goUL5zN58ni6dg0kIKAtwcHT8PAoaHQ8sRO77cGJjo4mMTGRkJAQBg0axJQpU1Lvi42NZfr06Xz88cesW7eO4sWLc/36dXr16sXKlStZuXIlTzzxBFOnTgXg+PHjTJ06NfU+lRsREXkUjo6O9O37Ftu3f0WVKl6Eha2lTp1X2LbtC6OjiZ3YreAcPnwYb29vACpXrszx48dT7zty5AjlypVj6tSpBAYGUrhw4dS9MwBffvkl7u7uvPrqqwCcOHGCRYsW0aFDBxYuXGivyCIiYnLlypVn06ZoRo4cw7VrV+nYsS3vvNOXmzf/MjqaZDC7HaKKjY3F1dU19bajoyPJyck4OTlx/fp1Dh48SEREBHnz5qVjx45UrlyZZ555BoCFCxcyc+bM1Oc2b96cwMBAXF1d6d+/Pzt37qRevXoP3baHR16cnBzt9dLw9HSz27ol82me5qFZmos95zlx4ljatQugS5curF79GXv27GLJkiW89tprdttmTmbEe9NuBcfV1ZW4uLjU21arFSenu5srUKAAL7zwAp6engBUrVqVkydP8swzz/Djjz/i7u6eer6OzWajS5cuuLnd/c3x8fHhhx9++MeCc/16vL1eFp6ebly5cstu65fMpXmah2ZpLpkxzyeeKMXGjdHMmvUBH344ncaNG9O58xuMHTsBV1eV5Yxi71k+rDzZ7RCVl5cXMTExABw9epRy5cql3lexYkXOnDnDtWvXSE5O5rvvvqNs2bIA7Nu3jzp16qQ+NjY2lhYtWhAXF4fNZuPgwYNUqlTJXrFFRCQHyZUrF0OGvMcXX+ygQoXnWbHiU+rWrcXevXuMjib/kt0KTqNGjXB2dqZ9+/ZMnjyZ9957j6VLl7J9+3YKFSrEoEGD6NGjB23btqVRo0apBejnn3/m6aefTl2Pm5sbAwcOpHPnzgQGBlK2bFl8fHzsFVtERHKgF1+szJdf7mbAgMFcuPArrVs3Z8SIIfcciZDsxWIz4YU67L0rTLvBzUPzNA/N0lyMnOe33x7irbd685//nOGZZ0oze/bH1KjxiiFZzMB0h6hERESyIy+vqmzf/hV9+77NuXM/06pVY8aMGcnt27eNjiaPQAVHRETkf+TOnZuxYycSFbWVUqWeYcGCOTRs6M3hw98YHU3SSQVHRETkIWrUeIWdO/fRs+fdQ1bNmzdi0qRxJCQkGB1N0qCCIyIi8g/y5s3LpEnTWL9+E0899TQffTSD117z4dixo0ZHk3+ggiMiIpIOtWt7s2vXfrp06c7Jkz/QpEl9pk0LJikpyeho8gAqOCIiIunk6urK9OkfsnZtBEWLPsEHH0yhSZP6/PDDCaOjyf9QwREREXlEdevWZ/fu/QQGBvH999/RqFEdZs36gOTkZKOjyf+ngiMiIvIY3N3zM2vWPFatWkvBgoUIDh5P8+YNOXPmtNHRBBUcERGRf6VRoybs2XOQ119vx5Ej39KgwavMmzeblJQUo6PlaCo4IiIi/1KBAh7Mn7+YZcs+x83NnXHjRtGqVRN++ulHo6PlWCo4IiIiGaRZsxbs2fM1vr7+fPPNQerVq83ixQuwWq1GR8txVHBEREQyUKFChVi8eBmLFy8jT548jBw5DH//Fpw/f87oaDmKCo6IiIgd+Pr6ExPzNU2btmDfvq/w8anJsmVLMOE1rrMkFRwRERE7KVKkCMuWrWLevEXkypWLoUMH0qaNHxcu/Gp0NNNTwREREbEji8VCmzbt2bPnIA0bvkZMzE58fGry+ecrtTfHjlRwREREMsETTxRj1ap1fPTRfAAGDOhHx45t+OOP3w1OZk4qOCIiIpnEYrHQoUMndu/ej49PPaKjv8Tbuwbr1q3R3pwMpoIjIiKSyZ566mnWro1g+vRZJCUl0a9fL7p27cjly5eNjmYaKjgiIiIGsFgsdOnyBrt376dWrVfZsmUjdepUJzIy3OhopqCCIyIiYqCSJUsRHr6RSZOmcvv2bXr27EqPHl24evWq0dGyNRUcERERgzk4ONCzZx927txLtWo1iIpaj7d3dTZv3mh0tGxLBUdERCSLKF26LFFRXzBmzERu3bpJ166B9OnTgxs3rhsdLdtRwREREclCHB0d6dfvbbZv/4oqVbwIC1uLt3cNtm37wuho2YoKjoiISBZUrlx5Nm2KZsSI0Vy7dpWOHdvyzjt9uXnzL6OjZQsqOCIiIlmUk5MTAwYMZtu2GF544SVWr/4MH5+a7Nq1w+hoWZ4KjoiISBb3/PMV+eKLHQwePJxLl/6gbVs/Bg8eQGzsLaOjZVkqOCIiItnA3Yt1juCLL3ZQocLzrFjxKXXr1mLv3j1GR8uSVHBERESykRdfrMyXX+7mnXcGceHCr7Ru3ZwRI4YQFxdndLQsRQVHREQkm3FxcWHkyDFs3hzNs8+W45NPFlK/fm0OHjxgdLQsQwVHREQkm/Lyqkp09B769n2bc+d+plWrxowZM5Lbt28bHc1wKjgiIiLZWJ48eRg7diJRUVspVeoZFiyYQ8OG3nz77SGjoxlKBUdERMQEatR4hR079tKzZ2/+858zNGvWkODg8SQkJBgdzRAqOCIiIiaRL18+Jk2axvr1m3jqqaeZNesDXnvNh2PHjhodLdOp4IiIiJhM7dre7Nq1jy5dunPy5A80aVKfadOCSUpKMjpaplHBERERMSFXVzemT/+QtWsjKFKkKB98MIUmTerzww8njI6WKVRwRERETKxu3frExBwgMDCI77//jkaN6jBr1gckJycbHc2u7FZwrFYro0ePpl27dgQFBXH+/Pl77t+9ezdt27alTZs2jB07FpvNhs1mw9vbm6CgIIKCgpgxYwYAO3bsICAggHbt2rF27Vp7RRYRETEld/f8zJo1j1Wr1lKwYCGCg8fTokUjzpw5bXQ0u3Gy14qjo6NJTEwkJCSEo0ePMmXKFBYsWABAbGws06dPZ8WKFRQsWJDFixdz/fp1bt26RcWKFfn4449T15OUlMTkyZMJDQ0lT548dOjQgfr161O4cGF7RRcRETGlRo2aEBNzgJEjhxEaGkKDBq8yfPj79O7dD0dHR6PjZSi77cE5fPgw3t7eAFSuXJnjx4+n3nfkyBHKlSvH1KlTCQwMpHDhwhQsWJATJ05w6dIlgoKC6NmzJz/99BNnz56lRIkS5M+fH2dnZ15++WW++eYbe8UWERExNQ+Pgsyfv5ilS1fh5ubOuHGjaNWqCT/99KPR0TKU3fbgxMbG4urqmnrb0dGR5ORknJycuH79OgcPHiQiIoK8efPSsWNHKleujKenJ7169aJp06YcOnSIIUOG8N577+Hm5pa6nnz58hEbG/uP2/bwyIuTk/2aqKenW9oPkmxD8zQPzdJcNE/76to1kBYtXqNv376sW7eOevVqM2XKFPr374+DQ8bu/zBilnYrOK6urvdc+MtqteLkdHdzBQoU4IUXXsDT0xOAqlWrcvLkSerVq5e6i6xq1apcvnz5vvXExcXdU3ge5Pr1+Ix+Oak8Pd24ckWXpzcLzdM8NEtz0Twziwvz5i3htdeaM2zYu7zzzjuEhKzjo4/mU7JkqQzZgr1n+bDyZLdDVF5eXsTExABw9OhRypUrl3pfxYoVOXPmDNeuXSM5OZnvvvuOsmXLMnfuXJYvXw7AqVOnKFasGGXKlOH8+fPcuHGDxMREDh06RJUqVewVW0REJMfx9fUnJuZrmjRpzr59X+HjU5Nly5Zgs9mMjvbYLDY7pbdarYwdO5YzZ85gs9kIDg4mJiaGEiVK0KBBAzZt2sSSJUsAaNKkCb169eKvv/5iyJAhxMfH4+joyOjRoylTpgw7duxg3rx52Gw2AgIC6Nix4z9u295NUf+qMA/N0zw0S3PRPI1hs9kIDQ1hxIih/PXXDXx86jFr1jyKF3/qsddp1B4cuxUcI6ngSHppnuahWZqL5mmsP/74nXfffYvo6C9xc3NnwoTJdOjQCYvF8sjrMt0hKhEREcmenniiGKtWrWPWrLtHTwYM6EfHjm3444/fjY6Wbio4IiIich+LxUJgYBAxMQeoU6ce0dFf4u1dg3Xr1mSLc3NUcEREROShnnrqadati2DatA9JSkqiX79edO3akcuXLxsd7R+p4IiIiMg/slgsdO3anV279lGr1qts2bKROnWqExkZbnS0h1LBERERkXQpVeoZwsM3MmnSVG7fvk3Pnl3p2bMrV69eNTrafVRwREREJN0cHBzo2bMPO3Z8RbVqNYiMDMfbuzqbN280Oto9VHBERETkkZUp8yxRUV8wZsxEbt26SdeugfTt25MbN64bHQ1QwREREZHH5OjoSL9+bxMdvYcqVbwIDQ3B27sG27Z9YXQ0FRwRERH5d8qXf45Nm6IZMWI0165dpWPHtrzzTl9u3vzLsEwqOCIiIvKvOTk5MWDAYLZti+GFF15i9erP8PGpybZt2wzJo4IjIiIiGeb55yvyxRc7GDx4OJcu/cFrr73GkSOHMz2HU6ZvUUREREwtV65cDB06giZNmrFxYzglSpTK9AwqOCIiImIXL75YmQYNvA25cKoOUYmIiIjpqOCIiIiI6ajgiIiIiOmo4IiIiIjpqOCIiIiI6ajgiIiIiOmo4IiIiIjpqOCIiIiI6ajgiIiIiOmo4IiIiIjpqOCIiIiI6ajgiIiIiOmo4IiIiIjpWGw2m83oECIiIiIZSXtwRERExHRUcERERMR0VHBERETEdFRwRERExHRUcERERMR0VHBERETEdFRwRERExHRyZMGxWq2MHj2adu3aERQUxPnz5++5f+3atfj7+9O2bVt27twJwLVr13jjjTcIDAxkwIAB3L59+6GP/duyZcv44IMPMudFCZCxs/37vsaNG5OQkJCpr0Pu9Thz/Zveh1lXWnMFvQezo++++46goKD7lu/YsYOAgADatWvH2rVr7R/ElgNt3brVNmzYMJvNZrMdOXLE1rt379T7Ll++bGvRooUtISHBdvPmzdSfJ0yYYAsLC7PZbDbbwoULbUuXLn3oY2/fvm179913bY0aNbJNnz7dkNeYU2XUbG02my0mJsbm6+trq1Kliu3OnTuZ/lrk/zzOXPU+zPr+aa42m96D2dGiRYtsLVq0sLVp0+ae5YmJibaGDRvabty4YUtISLD5+/vbrly5YtcsOXIPzuHDh/H29gagcuXKHD9+PPW+Y8eOUaVKFZydnXFzc6NEiRKcOnXqnufUqVOHffv2PfSxCQkJtG7dmt69exvy+nKyjJotgIODA0uXLqVAgQKZ/jrkXo8zV70Ps75/mivoPZgdlShRgjlz5ty3/OzZs5QoUYL8+fPj7OzMyy+/zDfffGPXLDmy4MTGxuLq6pp629HRkeTk5NT73NzcUu/Lly8fsbGx9yzPly8ft27deuhj8+fPz6uvvppJr0b+W0bNFqB27dp4eHhkYnp5mMeZq96HWd8/zRX0HsyOGjdujJOT033LH/Y+taccWXBcXV2Ji4tLvW21WlMH8r/3xcXF4ebmds/yuLg43N3dH/pYMU5GzVaylseZq2R9/zRXMRcj3qc5suB4eXkRExMDwNGjRylXrlzqfS+++CKHDx8mISGBW7ducfbsWcqVK4eXlxe7d+8GICYmhpdffvmhjxXjZNRsJWt5nLlK1vdPcxVzKVOmDOfPn+fGjRskJiZy6NAhqlSpYtdt5siq3KhRI/bu3Uv79u2x2WwEBwezdOlSSpQoQYMGDQgKCiIwMBCbzcbAgQNxcXGhT58+DBs2jLVr1+Lh4cGMGTPImzfvAx8rxsmo2UrW8jhzlawvrblK9rdhwwbi4+Np164dw4cPp3v37thsNgICAihatKhdt22x2Ww2u25BREREJJPlyENUIiIiYm4qOCIiImI6KjgiIiJiOio4IiIiYjoqOCIiImI6OfJj4iJiHrGxscyYMYNvvvkGR0dH3N3dGT58OBUrVjQ6mogYSAVHRLItq9VKz549qVGjBhERETg5OXHgwAF69uzJpk2b9DX/IjmYCo6IZFsHDx7k8uXLvP322zg43D3i/sorrzB58mSsVqvB6UTESCo4IpJt/fDDD7zwwgup5eZvPj4+BiUSkaxCJxmLSLbl4OCAvoxdRB5EBUdEsq1KlSrxww8/3FdyZs6cyYEDBwxKJSJZgQqOiGRbVatWpVChQsydO5eUlBQA9uzZQ3h4OGXLljU4nYgYSRfbFJFs7dq1a0yePJnjx4/j5OSEh4cHw4cP5/nnnzc6mogYSAVHRERETEeHqERERMR0VHBERETEdFRwRERExHRUcERERMR0VHBERETEdFRwRERExHRUcERERMR0/h8tJa56cGKVYgAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 576x396 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "analyser.plot_results(type=\"accuracy+fairness\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Accuracy and Fairness on test data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__[Model 6]: Best Fairness__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test set: 0.7166666666666667\n",
      "Fairness (True Positive Rate Difference) on test set: -0.0047916666666666385\n"
     ]
    }
   ],
   "source": [
    "# Best accuracy+fairness\n",
    "test_accuracy = analysis['best_accuracy_fairness_model'].get_accuracy(X_test, y_test)\n",
    "test_fairness = analysis['best_accuracy_fairness_model'].get_fairness(selected_dataset.dataset_orig_test)\n",
    "\n",
    "print(\"Accuracy on test set: \" + str(test_accuracy))\n",
    "print(\"Fairness (True Positive Rate Difference) on test set: \" + str(test_fairness['eq_opp_diff']))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "1893b06c1eb306a1e719abae99dfb3b9acc49f08e12646795ff8c000a93fc33e"
  },
  "kernelspec": {
   "display_name": "Python 3.9.12 ('ml')",
   "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": "3.9.12"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}