code/models/xgboost_pgd.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# XGBoost with PGD Dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Imports and Loading Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import xgboost as xgb\n",
    "import joblib\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load adversarial data\n",
    "adv_train = pd.read_csv(\"/Users/aryankargwal/adverscredit/models/data/pgd_train.csv\")\n",
    "adv_test = pd.read_csv(\"/Users/aryankargwal/adverscredit/models/data/pgd_test.csv\")\n",
    "\n",
    "# Assuming the last column is the label and the rest are features\n",
    "x_train_adv = adv_train.iloc[:, :-1]  # Features\n",
    "y_train_adv = adv_train.iloc[:, -1]   # Labels\n",
    "x_test_adv = adv_test.iloc[:, :-1]    # Features\n",
    "y_test_adv = adv_test.iloc[:, -1]     # Labels"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Defining XGBoost and Parameters for Grid Search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 32 candidates, totalling 160 fits\n"
     ]
    }
   ],
   "source": [
    "# Define XGBoost parameters to search over\n",
    "param_grid = {\n",
    "    'n_estimators': [100, 200],\n",
    "    'max_depth': [3, 4],\n",
    "    'learning_rate': [0.01, 0.1],\n",
    "    'subsample': [0.8, 0.9],\n",
    "    'colsample_bytree': [0.8, 1.0]\n",
    "}\n",
    "\n",
    "# Initialize the XGBoost classifier\n",
    "xgb_model_adv = xgb.XGBClassifier(eval_metric='mlogloss')\n",
    "\n",
    "# Set up GridSearchCV\n",
    "grid_search_adv = GridSearchCV(estimator=xgb_model_adv, param_grid=param_grid, cv=5, scoring='accuracy', verbose=1, n_jobs=-1)\n",
    "\n",
    "# Fit GridSearchCV with adversarial data\n",
    "grid_search_adv.fit(x_train_adv, y_train_adv)\n",
    "\n",
    "# Get the best model\n",
    "best_model_adv = grid_search_adv.best_estimator_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification Report (Adversarial Data):\n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "         0.0       1.00      1.00      1.00     56750\n",
      "         1.0       1.00      1.00      1.00     56976\n",
      "\n",
      "    accuracy                           1.00    113726\n",
      "   macro avg       1.00      1.00      1.00    113726\n",
      "weighted avg       1.00      1.00      1.00    113726\n",
      "\n",
      "Best adversarial model saved successfully.\n"
     ]
    }
   ],
   "source": [
    "# Evaluate the best model on adversarial test data\n",
    "y_pred_adv = best_model_adv.predict(x_test_adv)\n",
    "print(\"Classification Report (Adversarial Data):\\n\", classification_report(y_test_adv, y_pred_adv))\n",
    "\n",
    "# Save the best model\n",
    "joblib.dump(best_model_adv, 'best_xgboost_adversarial_model.pkl')\n",
    "\n",
    "print(\"Best adversarial model saved successfully.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluating Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification Report (Adversarial Data):\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0       1.00      1.00      1.00     56750\n",
      "         1.0       1.00      1.00      1.00     56976\n",
      "\n",
      "    accuracy                           1.00    113726\n",
      "   macro avg       1.00      1.00      1.00    113726\n",
      "weighted avg       1.00      1.00      1.00    113726\n",
      "\n",
      "Confusion Matrix (Adversarial Data):\n",
      "[[56688    62]\n",
      " [   83 56893]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0YAAAJuCAYAAACUrBL3AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8fJSN1AAAACXBIWXMAAA9hAAAPYQGoP6dpAAA/hElEQVR4nO3debyXc+L//+epTqc9SgsRYixlq2wZydooE5mxja1kX2bsjPERYwtjxpIlS4nG2GmyTGNfZmwxZIuQiqmkTJYkqffvD9/Oz1FRVCeu+/12O7db53pfy+u6zonz6Lrer1NWKpVKAQAAKLAa1T0AAACA6iaMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4yAZdJLL72UAw44IKuvvnrq1KmTBg0apEOHDrngggvy4YcfLtFjv/DCC+nSpUsaN26csrKyXHzxxYv9GGVlZTnjjDMW+36/y+DBg1NWVpaysrI8+uij87xeKpWy5pprpqysLFtvvfX3OsYVV1yRwYMHL9I2jz766ALH9EOceeaZadu2bebMmTPPa1OmTElFRUXKysry3HPPLdJ+q+vrtzSdccYZKSsrW6Lbzl1v7ke9evWy8sor5xe/+EX69++fTz755HsdP0mefPLJnHHGGZk2bdr33sdpp52WDh06zPf7B/jpEUbAMueaa65Jx44dM2LEiJx44okZPnx47rrrruy+++4ZMGBADjzwwCV6/D59+mTixIm5+eab89RTT2WvvfZa7Md46qmnctBBBy32/S6shg0bZuDAgfMsf+yxx/L222+nYcOG33vf3yeMOnTokKeeeiodOnT43sf9pgkTJuSCCy7ImWeemRo15v3f3ZAhQ/LFF18kyXyvRdEddNBBeeqpp5bKsYYPH56nnnoqw4cPz4UXXpjWrVvnpJNOSrt27TJy5Mjvtc8nn3wyf/zjH39QGJ1wwgl55513cv3113/vfQA/HrWqewAAX/fUU0/l8MMPzw477JChQ4emoqKi8rUddtghxx9/fIYPH75Ex/DKK6/k4IMPTrdu3ZbYMTbffPMltu+Fseeee+bGG2/M5ZdfnkaNGlUuHzhwYDp16pSPP/54qYxj1qxZKSsrS6NGjRb7Nbnkkkuy3HLL5Ve/+tV8Xx80aFCaN2+eVVddNTfddFP+8pe/pG7duot1DIvTjBkzUqdOne99F2dhffbZZ5V3blZeeeUleqy5OnbsmBVWWKHy87322itHHXVUunTpkp133jmjR4+u8t+CpaVx48bZd999c95556V3795L/NoD1csdI2CZcu6556asrCxXX331fH8Qql27dnbeeefKz+fMmZMLLrgg66yzTioqKtK8efPsv//+ee+996pst/XWW2e99dbLiBEj0rlz59SrVy9t2rTJeeedV/mYzNzHzL788stceeWVlY/3JAt+NGjuNmPHjq1c9vDDD2frrbdO06ZNU7du3bRu3Tq//vWv89lnn1WuM79HsV555ZXssssuWX755VOnTp1stNFG8/xL9dxHzm666aaceuqpWWmlldKoUaNsv/32eeONNxbuIif5zW9+kyS56aabKpd99NFHueOOO9KnT5/5bvPHP/4xm222WZo0aZJGjRqlQ4cOGThwYEqlUuU6q622Wl599dU89thjlddvtdVWqzL2IUOG5Pjjj0+rVq1SUVGRt956a55H6aZMmZJVVlklW2yxRWbNmlW5/9deey3169fPfvvt963n98UXX2TgwIHZe++953u36Jlnnskrr7yS/fbbLwcffHDluX/Txx9/nIMPPjhNmzZNgwYNsuOOO2b06NFV1hk6dGjKysry0EMPzbP93O+jl156qXLZc889l5133jlNmjRJnTp10r59+9x6661Vtpv7fXX//fenT58+adasWerVq5eZM2fmgw8+yCGHHJJVVlklFRUVadasWX7+85/nwQcfrNz+gQceyC677JKVV145derUyZprrplDDz00U6ZMqXKcud/X//nPf7Lbbrtl+eWXzxprrFHlta+75ZZb0rVr16y44oqpW7du1l133fz+97/P9OnTF/Sl+N423HDDnHrqqRk/fnxuueWWRTq3M844IyeeeGKSZPXVV5/n8dFFOY/99tsvo0ePziOPPLLYzxFYtggjYJkxe/bsPPzww+nYsWNWWWWVhdrm8MMPz8knn5wddtghw4YNy1lnnZXhw4dniy22mOeHwEmTJmWfffbJvvvum2HDhqVbt2455ZRT8te//jVJstNOO1U+OrTbbrvlqaeeWuRHicaOHZuddtoptWvXzqBBgzJ8+PCcd955qV+/fuVjW/PzxhtvZIsttsirr76aSy+9NHfeeWfatm2b3r1754ILLphn/T/84Q8ZN25crr322lx99dV5880306NHj8yePXuhxtmoUaPstttuGTRoUOWym266KTVq1Miee+65wHM79NBDc+utt+bOO+/Mr371q/z2t7/NWWedVbnOXXfdlTZt2qR9+/aV1++uu+6qsp9TTjkl48ePz4ABA3L33XenefPm8xxrhRVWyM0335wRI0bk5JNPTvLVnYzdd989rVu3zoABA771/J555plMnTo122yzzXxfn/voXJ8+fbLXXnulXr168zxOVyqV0rNnz8qQu+uuu7L55pvPcyfxl7/8ZZo3b57rrrtunuMMHjw4HTp0yAYbbJAkeeSRR/Lzn/8806ZNy4ABA/L3v/89G220Ufbcc8/5Pn7Yp0+flJeXZ8iQIbn99ttTXl6e/fbbL0OHDk3fvn1z//3359prr83222+fqVOnVm739ttvp1OnTrnyyitz//33p2/fvnnmmWey5ZZbVgnNuX71q19lzTXXzG233fat1/bNN99M9+7dM3DgwAwfPjzHHHNMbr311vTo0WOB2/wQc/8R5PHHH69ctjDndtBBB+W3v/1tkuTOO++s/F6c+6jmopxHx44d06BBg9x7771L5ByBZUgJYBkxadKkUpLSXnvttVDrjxo1qpSkdMQRR1RZ/swzz5SSlP7whz9ULuvSpUspSemZZ56psm7btm1Lv/jFL6osS1I68sgjqyw7/fTTS/P7T+Z1111XSlJ65513SqVSqXT77beXkpRefPHFbx17ktLpp59e+flee+1VqqioKI0fP77Ket26dSvVq1evNG3atFKpVCo98sgjpSSl7t27V1nv1ltvLSUpPfXUU9963LnjHTFiROW+XnnllVKpVCptsskmpd69e5dKpVKpXbt2pS5duixwP7Nnzy7NmjWrdOaZZ5aaNm1amjNnTuVrC9p27vG22mqrBb72yCOPVFl+/vnnl5KU7rrrrlKvXr1KdevWLb300kvfeo5f327SpEnzvDZ9+vRSo0aNSptvvnnlsl69epXKyspKb731VuWyf/zjH6UkpUsuuaTK9uecc848X7/jjjuuVLdu3cqvU6lUKr322mulJKX+/ftXLltnnXVK7du3L82aNavKPn/5y1+WVlxxxdLs2bNLpdL//3Xaf//95xl/gwYNSsccc8x3XoO55syZU5o1a1Zp3LhxpSSlv//975Wvzf2+7tu37zzbLeh7/pv7feyxx0pJSiNHjlzobb+53gcffDDf12fMmFFKUurWrdsin9uf/vSnKn83v895zPXzn/+8tNlmm33n+QA/bu4YAT9acx9t6d27d5Xlm266adZdd915Hm1q2bJlNt100yrLNthgg4wbN26xjWmjjTZK7dq1c8ghh+T666/PmDFjFmq7hx9+ONttt908d8p69+6dzz77bJ47V19/nDBJ5R2JRTmXLl26ZI011sigQYPy8ssvZ8SIEQt8jG7uGLfffvs0btw4NWvWTHl5efr27ZupU6dm8uTJC33cX//61wu97oknnpiddtopv/nNb3L99denf//+WX/99b9zuwkTJqSsrKzK+1bmuvXWW/Pxxx9XOdc+ffqkVCpVuesz9/trn332qbL93nvvPc8++/TpkxkzZlR55Ou6665LRUVF5fpvvfVWXn/99cr9ffnll5Uf3bt3z8SJE+d5HHJ+12rTTTfN4MGDc/bZZ+fpp5+e7x2gyZMn57DDDssqq6ySWrVqpby8PKuuumqSZNSoUfOsv7BfkzFjxmTvvfdOy5YtK78HunTpssD9/lClrz2mOdeintv8LOp5NG/ePP/9739/wJkAPwbCCFhmrLDCCqlXr17eeeedhVp/7qNDK6644jyvrbTSSlUeLUqSpk2bzrNeRUVFZsyY8T1GO39rrLFGHnzwwTRv3jxHHnlk1lhjjayxxhq55JJLvnW7qVOnLvA85r7+dd88l7nvx1qUcykrK8sBBxyQv/71rxkwYEDWWmutdO7ceb7rPvvss+natWuSr2YN/Pe//50RI0bk1FNPXeTjzu88v22MvXv3zueff56WLVt+53uL5poxY0bKy8tTs2bNeV4bOHBg6tSpkx133DHTpk3LtGnTssEGG2S11VbL4MGDKx9HnDp1amrVqjXPtW7ZsuU8+2zXrl022WSTyrCaPXt2/vrXv2aXXXZJkyZNkiTvv/9+kq9mOisvL6/yccQRRyTJPI9/zu9a3XLLLenVq1euvfbadOrUKU2aNMn++++fSZMmJfnqfXddu3bNnXfemZNOOikPPfRQnn322Tz99NOV1+abFuZr8umnn6Zz58555plncvbZZ+fRRx/NiBEjcueddy5wvz/U3NCf+/fg+5zb4jiPOnXqLJHzA5YtZqUDlhk1a9bMdtttl3/84x957733vnNGrLk/sE6cOHGedSdMmDDfuwXfV506dZIkM2fOrDIpxDd/kE2Szp07p3Pnzpk9e3aee+659O/fP8ccc0xatGixwKm/mzZtmokTJ86zfMKECUmyWM/l63r37p2+fftmwIABOeeccxa43s0335zy8vLcc889ldci+WrigUW1KDN7TZw4MUceeWQ22mijvPrqqznhhBNy6aWXfud2K6ywQr744otMnz499evXr1w+evTo/Otf/0qStG7der7b/vOf/0z37t3TtGnTfPnll5k6dWqVOJobIN90wAEH5IgjjsioUaMyZsyYTJw4MQcccECVMSVfvcdqQTPlrb322lU+n9+1WmGFFXLxxRfn4osvzvjx4zNs2LD8/ve/z+TJkzN8+PC88sorGTlyZAYPHpxevXpVbvfWW2/N95gLOs43Pfzww5kwYUIeffTRyrsrSX7QdNjfZdiwYUlS+Tu1vs+5fdP3OY8PP/xwif0dBJYd7hgBy5RTTjklpVIpBx988HwnK5g1a1buvvvuJMm2226bJJWTJ8w1YsSIjBo1Ktttt91iG9fcmdW+PrtYksqxzE/NmjWz2Wab5fLLL0+S/Oc//1ngutttt13lD2xfd8MNN6RevXpLbHrvVq1a5cQTT0yPHj2q/KD5TWVlZalVq1aVOzAzZszIkCFD5ll3cd2Fmz17dn7zm9+krKws//jHP9KvX7/079+/8l/2v80666yT5Ks36n/d3AkWrrnmmjzyyCNVPu67776Ul5dXTkgxd+KGG2+8sco+/va3v833mL/5zW9Sp06dDB48OIMHD06rVq0q77IlX0XPz372s4wcOTIbb7zxfD8W9fdHtW7dOkcddVR22GGHyu+vuZHzzVkdr7rqqkXa9zctqf0uyMiRI3PuuedmtdVWyx577LHIY1jQXdTvcx5jxoxJ27ZtF/EMgB8bd4yAZcrc2aaOOOKIdOzYMYcffnjatWuXWbNm5YUXXsjVV1+d9dZbLz169Mjaa6+dQw45JP3790+NGjXSrVu3jB07NqeddlpWWWWVHHvssYttXN27d0+TJk1y4IEH5swzz0ytWrUyePDgvPvuu1XWGzBgQB5++OHstNNOad26dT7//PPKH7S33377Be7/9NNPzz333JNtttkmffv2TZMmTXLjjTfm3nvvzQUXXJDGjRsvtnP5pvPOO+8719lpp53yl7/8JXvvvXcOOeSQTJ06NRdeeOF8p1Rff/31c/PNN+eWW25JmzZtUqdOnYV6X9A3nX766XniiSdy//33p2XLljn++OPz2GOP5cADD0z79u2z+uqrL3DbuXcYnn766cr3X3355Ze54YYbsu666y7wl+v26NEjw4YNywcffJCuXbtmq622ykknnZTp06dn4403zr///e/5xmCSLLfcctl1110zePDgTJs2LSeccMI8U4VfddVV6datW37xi1+kd+/eadWqVT788MOMGjUq//nPf3Lbbbd96zX56KOPss0222TvvffOOuusk4YNG2bEiBEZPnx45V2oddZZJ2ussUZ+//vfp1QqpUmTJrn77rvzwAMPfOu+v8sWW2yR5ZdfPocddlhOP/30lJeX58Ybb/zev4D1655//vk0btw4s2bNyoQJE/LQQw9lyJAhad68ee6+++7Url17kc9t7vfcJZdckl69eqW8vDxrr732Ip/H1KlT8+abb1bOcgf8hFXr1A8AC/Diiy+WevXqVWrdunWpdu3apfr165fat29f6tu3b2ny5MmV682ePbt0/vnnl9Zaa61SeXl5aYUVVijtu+++pXfffbfK/rp06VJq167dPMfp1atXadVVV62yLPOZla5UKpWeffbZ0hZbbFGqX79+qVWrVqXTTz+9dO2111aZ+eqpp54q7brrrqVVV121VFFRUWratGmpS5cupWHDhs1zjK/PalYqlUovv/xyqUePHqXGjRuXateuXdpwww1L1113XZV15s7edtttt1VZ/s4775SSzLP+N319VrpvM7+Z5QYNGlRae+21SxUVFaU2bdqU+vXrVxo4cOA8M3+NHTu21LVr11LDhg1LSSqv74LG/vXX5s5Kd//995dq1KgxzzWaOnVqqXXr1qVNNtmkNHPmzG89h86dO1eZvW/o0KGlJKWLL754gdsMHz68lKT05z//uVQqlUrTpk0r9enTp7TccsuV6tWrV9phhx1Kr7/++ny/fnPHnaSUpDR69Oj5HmPkyJGlPfbYo9S8efNSeXl5qWXLlqVtt922NGDAgMp1FvR1+vzzz0uHHXZYaYMNNig1atSoVLdu3dLaa69dOv3000vTp0+vXO+1114r7bDDDqWGDRuWll9++dLuu+9eGj9+/Dzj/rZZ4eY3s9yTTz5Z6tSpU6levXqlZs2alQ466KDSf/7zn3m+9xZ1Vrq5HxUVFaUVV1yx1LVr19Ill1xS+vjjj+fZZmHPrVQqlU455ZTSSiutVKpRo0aV76+FPY9SqVQaOHBgqby8fL4zHAI/LWWl0nymfAGAH7k77rgje+65Z8aNG5dWrVpV93D4kercuXNat249zyOVwE+PMALgJ6lUKmWLLbZIx44dc9lll1X3cPgRevzxx9O1a9e89tpradOmTXUPB1jCTL4AwE9SWVlZrrnmmqy00kqZM2dOdQ+HH6GpU6fmhhtuEEVQEO4YAQAAheeOEQAAUHjCCAAAKDxhBAAAFJ4wAgAACq9WdQ9gSajb/qjqHgIAy4D/jTBNN0DR1VnI4nHHCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeLWqewBAcuqh3fN/h3WvsmzSlI+z+g5/qPx87dVb5Oyje6ZzhzVTo0ZZRr09MfuePCjvTvpf5TqbbbB6zjjyl9lk/dUy68vZeemN/2aXo67I5zNnJUnWbN085x7bM502bJPa5TXz6lsTcsbl9+Tx596s3EfHtq1z1u92Sfu2q6RUSp5/dVxOvXhoXhr93yV8FQD4vt5///1c/Jc/5d9PPJGZMz/PqquuljPOOidt262XWbNm5bJLL86/nng87733bho2aJDNOm2Ro489Ps2bt6juocMyQxjBMuLVtyZkp8P6V34+e06p8s+rr7xCHhp0XK4f+mTOvvLefPTpjKyzesvK4Em+iqK/X3ZELrzu/hx3/m354svZ2WCtVpnztf3c1f+wvDlucrodemlmzJyVo/beJndeelja9Tgj70/9JA3qVWTYFUfmnkdfztH9bkmtmjVy2uE7ZdgVR2bNHf8vX345Z+lcDAAW2scffZTe+/4mG2+6WS4fcE2aNG2S9959Nw0bNkqSfP7553l91Gs55LDDs/ba6+Tjjz/OBeedm6OPOjw33XpnNY8elh3CCJYRX86ek/enfjLf1/54VI/881+v5tRL/l65bOx/p1ZZ54Ljf5Urbn40F173QOWyt8d/UPnnpsvVz5qtm+ewM27MK29OSJKcdunfc9ieW2XdNVbM+1M/yVqrtUiTxvVz1pX35L33pyVJzrnqH3nutj9klZZN8s57UxbX6QKwmAwaeE1atGyZs87pV7msVauVK//csGHDXHXtdVW2+f0f/i/77LV7Jk6YkBVXWmmpjRWWZd5jBMuINVs3y5j7z8moe87IDecdkNVaNU2SlJWVZcct2+XN8ZMz7PIjM+6hfnn8hhPSY+sNKrdttnyDbLrB6vngw0/zyODjMvbBc3P/tUdni43aVK4zddr0jBozMXv/ctPUq1M7NWvWyEG/3jKTpnycF157N0kyeuz7+eB/n6RXzy1SXqtm6lSUp3fPTnn1rQkZP/HDpXtBAFgojz3ycNq1Wy8nHPu7bN25U/b4dc/ccdut37rNp59+mrKysjRs1GgpjRKWfWWlUqn03astGe+9916uvPLKPPnkk5k0aVLKysrSokWLbLHFFjnssMOyyiqrfOc+Zs6cmZkzZ1ZZ1rzzySmrUXNJDRsWu64/b5t6dWrnzXGT07xpw/z+oB2z1mot0nG3c1Jeq0bGPtgv02fMzB8vvyePjRidrj9vmz8e1SO/OOTS/Ov5t7Lp+qvlsRtOyNRp03PKRXflpTfeyz6/3DSH7NE5HXc/t/LO0UrNGufWiw9N+3VWzpw5pUz+8JPs+tsrq7x/aN02LXPbRYdWhtmb4yZn5yMvr/JeJvix+N+Iy6p7CLDEbdJ+/STJfr0OyA6/2DGvvPxS/nTeuTnt9DPTY5ee86w/c+bM9N7vN1lt9Tbpd/6FS3m0sPTVWchn5KrtjtG//vWvrLvuurnrrruy4YYbZv/998++++6bDTfcMEOHDk27du3y73//+zv3069fvzRu3LjKx5fvP78UzgAWn/v//VqGPvRiXn1rQh555o3s+tsrkyT79tgsNWp89df0nkdfTv8bH8lLo/+bC697IPc98WoO3m3LJEmNGmVJkoF3/CtDhj2dkW+8l5P+fGdGj52cXrt0qjzOxX/YMx98+Em273NxOu/3p9z96Eu589LD0nKFr/7FsE5Fea46Y988NXJMuux/YbY94C8ZNWZi7up/eOpUlC/NSwLAQpozp5R127bL7445Luuu2za777FXfrXbHrn1lpvmWXfWrFk5+YRjM2dOKaeedsbSHywsw6rtPUbHHntsDjrooFx00UULfP2YY47JiBEjvnU/p5xySo477rgqy5p3PnmxjROqw2eff5FX35qQNVo3y5T/fZpZs2Zn1JiJVdZ5Y8ykbNH+q0flJn7wcZJk1JhJVdd5Z1JWabl8kmTrTddK987rZcUuJ+WT6Z8nSY7pd2u223yd7Ntjs1x43QPZs9vGab1Sk3Tp9efMvZnc65TBmfj4Bemx9Qa57Z/+0QFgWdOsWbO0WWONKsvatGmTBx/4Z5Vls2bNyonHH5P/vvderrnu+jRo0GBpDhOWedV2x+iVV17JYYcdtsDXDz300LzyyivfuZ+Kioo0atSoyofH6Pixq11eK+us3iKTpnyUWV/OzvOvjctaq1adUvVnqzbP+IlfPd42bsLUTJg8LWut1rzKOmuu2rzyvUH16tROksyZU3VmuTlzSikrK6tcZ86cUr7+hO2cUimlUlLj/60DwLJlo/YdMvadd6osGzd2bFZaqVXl53OjaPy4cblq4OAst9zyS3uYsMyrtjBaccUV8+STTy7w9aeeeiorrrjiUhwRVJ9+x+6aLTuumVVXappN1ls1f/vTgWlYv05uvPuZJMlF1z+Y3X7RIQfsukXarLJCDttzq3Tfar1cfevjlfu46PoHc8ReW2fX7TdKm1VWSN8jdsraq7XI4KFPJUmeeemd/O/jz3LtWftn/bVaffU7jY7pmdVaNc3wf72aJHno6dezfKN6ufiUPbL26i2ybpuWufqMffPl7Nl57LnRS//CAPCd9t2/V15+aWSuvXpAxo8bl/vuuTu3335r9vzN3kmSL7/8Micc+7u89uor6Xf+hZkze3amfPBBpnzwQWZ98UU1jx6WHdU2+cIVV1yRY489NgcffHB22GGHtGjRImVlZZk0aVIeeOCBXHvttbn44ou/9a7SgtRtf9QSGDEsOTecd0C27LBmmi5XP1P+92mefXls/njFPXn9a4/G7b/L5jmxT9e0ar5cRo+bnLMH3Jt7Hn25yn5OOGCHHLrHVlm+cb28PPq/OfXioXnyxTGVr3do2zpnHNkjHdq2TnmtGhk1ZlLOvfofuf/fr1Wus+1m6+TUQ7ul7ZorZs6cUka+/l7OuPzuPPvy2CV+HWBxM/kCRfHYo4/k0ov/kvHjxqbVyitnv/0PyK933yNJ8t//vpfuXbeb73bXXndDNtl0s6U5VFjqFnbyhWqdle6WW27JRRddlOeffz6zZ89OktSsWTMdO3bMcccdlz322ON77VcYAZAIIwB+JGE016xZszJlyle/OHKFFVZIefkPm/1KGAGQCCMAFj6Mqm1Wuq8rLy/3fiIAAKDaVNvkCwAAAMsKYQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwqu1MCsNGzZsoXe48847f+/BAAAAVIeFCqOePXsu1M7Kysoye/bsHzIeAACApW6hwmjOnDlLehwAAADVxnuMAACAwluoO0bfNH369Dz22GMZP358vvjiiyqv/e53v1ssAwMAAFhaFjmMXnjhhXTv3j2fffZZpk+fniZNmmTKlCmpV69emjdvLowAAIAfnUV+lO7YY49Njx498uGHH6Zu3bp5+umnM27cuHTs2DEXXnjhkhgjAADAErXIYfTiiy/m+OOPT82aNVOzZs3MnDkzq6yySi644IL84Q9/WBJjBAAAWKIWOYzKy8tTVlaWJGnRokXGjx+fJGncuHHlnwEAAH5MFvk9Ru3bt89zzz2XtdZaK9tss0369u2bKVOmZMiQIVl//fWXxBgBAACWqEW+Y3TuuedmxRVXTJKcddZZadq0aQ4//PBMnjw5V1999WIfIAAAwJJWViqVStU9iMWtbvujqnsIACwD/jfisuoeAgDVrM5CPiPnF7wCAACFt8jvMVp99dUrJ1+YnzFjxvygAQEAACxtixxGxxxzTJXPZ82alRdeeCHDhw/PiSeeuLjGBQAAsNQschgdffTR811++eWX57nnnvvBAwIAAFjaFtt7jLp165Y77rhjce0OAABgqVlsYXT77benSZMmi2t3AAAAS833+gWvX598oVQqZdKkSfnggw9yxRVXLNbBfV+mZwUgSZbfxK9vACi6GS8sXBsschjtsssuVcKoRo0aadasWbbeeuuss846i7o7AACAaveT/AWvn39Z3SMAYFngjhEAC3vHaJHfY1SzZs1Mnjx5nuVTp05NzZo1F3V3AAAA1W6Rw2hBN5hmzpyZ2rVr/+ABAQAALG0L/R6jSy+9NElSVlaWa6+9Ng0aNKh8bfbs2Xn88ce9xwgAAPhRWugwuuiii5J8dcdowIABVR6bq127dlZbbbUMGDBg8Y8QAABgCVvoMHrnnXeSJNtss03uvPPOLL/88ktsUAAAAEvTIk/X/cgjjyyJcQAAAFSbRZ58Ybfddst55503z/I//elP2X333RfLoAAAAJamRQ6jxx57LDvttNM8y3fcccc8/vjji2VQAAAAS9Mih9Gnn34632m5y8vL8/HHHy+WQQEAACxNixxG6623Xm655ZZ5lt98881p27btYhkUAADA0rTIky+cdtpp+fWvf52333472267bZLkoYceyt/+9rfcfvvti32AAAAAS9oih9HOO++coUOH5txzz83tt9+eunXrZsMNN8zDDz+cRo0aLYkxAgAALFFlpVKp9EN2MG3atNx4440ZOHBgRo4cmdmzZy+usX1vn39Z3SMAYFmw/CZHVfcQAKhmM164bKHWW+T3GM318MMPZ999981KK62Uyy67LN27d89zzz33fXcHAABQbRbpUbr33nsvgwcPzqBBgzJ9+vTssccemTVrVu644w4TLwAAAD9aC33HqHv37mnbtm1ee+219O/fPxMmTEj//v2X5NgAAACWioW+Y3T//ffnd7/7XQ4//PD87Gc/W5JjAgAAWKoW+o7RE088kU8++SQbb7xxNttss1x22WX54IMPluTYAAAAloqFDqNOnTrlmmuuycSJE3PooYfm5ptvTqtWrTJnzpw88MAD+eSTT5bkOAEAAJaYHzRd9xtvvJGBAwdmyJAhmTZtWnbYYYcMGzZscY7vezFdNwCJ6boBWArTdSfJ2muvnQsuuCDvvfdebrrpph+yKwAAgGrzg3/B67LIHSMAEneMAFhKd4wAAAB+CoQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwalX3AIDv58svv8yAy/vn3nvvztQpU7JCs2bZeZddc8hhR6RGja/+zePKy/tn+D/uzaRJk1JeXp62bdvlqKOPzQYbbFjNowfgm049tHv+77DuVZZNmvJxVt/hD5Wfr716i5x9dM907rBmatQoy6i3J2bfkwfl3Un/S5K0aNow5x6za7bdfJ00rF+R0WMn50+D/pm7Hnyxch8brbNyzj66Zzq2a53Zs0sZ+tCLOfnPd2T6jC+SJE0a18915/TK+mu1SpPG9fLBh5/mnkdfSt/L7s4n0z9f8hcCqokwgh+p6wZek9tuvTlnnXt+1lhzzbz2yivp+3+npGHDhtlnv15JklVXXS2nnNo3K6+8Sj6f+Xn+esPgHH5wn9z9jwfSpEmTaj4DAL7p1bcmZKfD+ld+PntOqfLPq6+8Qh4adFyuH/pkzr7y3nz06Yyss3rLfD5zVuU6A8/ulcYN6mT3Y67KlGmfZs9uG2fIeX3y830uyMg33suKzRrn3gG/ze33/yfHnndrGtWvkz+d+Otcc+Z+2fvEgUmSOXPm5J7HXsofr7gnU/73Sdqs0iwX/36P9G9cP73/MHipXQtY2oQR/EiNHPlitt52u2zVZeskSatWK+cf992bV199pXKd7r/sUWWbE046JXfdcXveHP1GNtu809IcLgAL4cvZc/L+1E/m+9ofj+qRf/7r1Zx6yd8rl43979Qq62y2wer53bk357lXxyVJzr/2n/ntPttmo3VXycg33ku3zutl1pezc0y/W1MqfRVdx/S7Nc/cckrarLJCxrw7JdM+mZFrbvtX5T7HT/xfrr7tiRy7//aL+3RhmeI9RvAj1b59xzz79NMZO/adJMkbr7+eF154Pp07d5nv+rO++CJ33HZLGjZsmLXWXntpDhWAhbRm62YZc/85GXXPGbnhvAOyWqumSZKysrLsuGW7vDl+coZdfmTGPdQvj99wQnpsvUGV7Z984e3s1rVjlm9UL2VlZdn9Fx1TUbtWHn/uzSRJRe1amTVrdmUUJcmM/3fHaYuN1pjvmFZs1ji7bLtRnnj+zSVxyrDMWKbD6N13302fPn2+dZ2ZM2fm448/rvIxc+bMpTRCqD59Djo4O3bfKT1/2S0dN2yXPXfrmX3365VuO/2yynqPPfpINt+4fTbpsEGG3DA4A64ZlOWX9xgdwLJmxCtjc9BpQ9LjiMtzxFk3pUXTRnlk8PFp0rh+mjdpkIb16+SEA3bIA0++lh6HX5Zhj4zMzX8+KFt2XLNyH/v9flBq1ayRCY9dkI+euTj9T90rex53Td55b0qS5NFn30iLpo1y7P7bpbxWzSzXsG7O/O3OSZKWzRpXGc/1/Xpn6pN/yZj7z8nH0z/P4Wf+beldDKgGy3QYffjhh7n++uu/dZ1+/fqlcePGVT7+dH6/pTRCqD7D/3Ff7r1nWPpd8OfcfNudOevc83L9dYMybOhdVdbbZNPNcusdQ3PDjTfn51t2zonHH5OpU6cuYK8AVJf7//1ahj70Yl59a0IeeeaN7PrbK5Mk+/bYrHJSnXsefTn9b3wkL43+by687oHc98SrOXi3LSv3ccaRPbJ8o3rpduil+fm+F+TSvz6cG//UJ+3WXClJMmrMpBzcd0h+t992+fCpv2Tsg+fmnfemZNKUjzNn9pwq4znpwjvSae/zs/uxV6XNyivk/ON/tZSuBFSPan2P0bBhw7719TFjxnznPk455ZQcd9xxVZaValb8oHHBj8FFf74gfQ48JN2675Qk+dlaa2fihAkZeO1V2bnnrpXr1atXL61XXTWtV101G2y4UXp065qhd96eAw8+tLqGDsBC+OzzL/LqWxOyRutmmfK/TzNr1uyMGjOxyjpvjJmULdq3SfLV5AyH79UlHX59dkaNmZQkeXn0f/PzDmvk0D23yu/OuTlJcsvw53LL8OfSvEnDTJ8xM6VS8rt9t53n/UrvT/0k70/9JKPHvp8Pp03PQ9cdl/OuGZ5JUz5eCmcPS1+1hlHPnj1TVlZW5TnXbyorK/vWfVRUVKSiomoIff7lYhkeLNM+n/F5atSo+vejZs2amTNnwX+fkqRUKuWLL75YkkMDYDGoXV4r66zeIv9+4a3M+nJ2nn9tXNZatUWVdX62avOMn/jVVN316tROksz5xs9Vs2eXUmM+P09N/vCrSR7232XzfP7FrDz09OsLHMvcn8dql5u3i5+uav3uXnHFFXP55ZenZ8+e8339xRdfTMeOHZfuoOBHosvW2+Saqwek5YorZY0118zro0ZlyPXXZZddf50k+eyzz3Lt1QOy9TbbZoVmzfLRtGm55ea/5f33J2WHX+xYzaMH4Jv6Hbtr7n385bw78X9p3qRBTj5oxzSsXyc33v1MkuSi6x/MkPP75F//eSuPPTc6Xbdom+5brZdfHHxJkuSNsZPy1vjJuez/fpNT/nJXpn40PTtvs0G223zt/OroAZXHOWzPrfL0yDH59LMvst3m6+TcY3rmtP5/z0efzkiS/GLLtmnepFGef3VcPv1sZtZdo2XOObpnnnzh7Yyf+OHSvzCwlJSVvu12zRK28847Z6ONNsqZZ54539dHjhyZ9u3bZ86cOfN9fUHcMaIIpk//NJdfekkefujBfPjh1DRr3jzduu2UQw8/MuW1a2fmzJn5/UnH5+WXRmba//6X5ZZbLu3WWz8HH3p41lt/g+8+APwELL/JUdU9BFhoN5x3QLbssGaaLlc/U/73aZ59eWz+eMU9ef3/PRaXfHV358Q+XdOq+XIZPW5yzh5wb+559OXK19do3Sxn/26XdNqoTRrUq8jb736Qi294KDfdO6JynWvP2i87brleGtSrnTfGvj/P61tt/LP88ageWadNy1SU18p770/L3x9+MRcOeqAynuDHZMYLly3UetUaRk888USmT5+eHXec/79eT58+Pc8991y6dJn/9MMLIowASIQRAD+SMFpShBEAiTACYOHDaJmerhsAAGBpEEYAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIAAAoPGEEAAAUnjACAAAKTxgBAACFJ4wAAIDCE0YAAEDhCSMAAKDwhBEAAFB4wggAACg8YQQAABSeMAIAAApPGAEAAIVXViqVStU9CGDxmjlzZvr165dTTjklFRUV1T0cAKqJ/x/AwhNG8BP08ccfp3Hjxvnoo4/SqFGj6h4OANXE/w9g4XmUDgAAKDxhBAAAFJ4wAgAACk8YwU9QRUVFTj/9dG+0BSg4/z+AhWfyBQAAoPDcMQIAAApPGAEAAIUnjAAAgMITRgAAQOEJI/iJueKKK7L66qunTp066dixY5544onqHhIAS9njjz+eHj16ZKWVVkpZWVmGDh1a3UOCZZ4wgp+QW265Jcccc0xOPfXUvPDCC+ncuXO6deuW8ePHV/fQAFiKpk+fng033DCXXXZZdQ8FfjRM1w0/IZtttlk6dOiQK6+8snLZuuuum549e6Zfv37VODIAqktZWVnuuuuu9OzZs7qHAss0d4zgJ+KLL77I888/n65du1ZZ3rVr1zz55JPVNCoAgB8HYQQ/EVOmTMns2bPTokWLKstbtGiRSZMmVdOoAAB+HIQR/MSUlZVV+bxUKs2zDACAqoQR/ESssMIKqVmz5jx3hyZPnjzPXSQAAKoSRvATUbt27XTs2DEPPPBAleUPPPBAtthii2oaFQDAj0Ot6h4AsPgcd9xx2W+//bLxxhunU6dOufrqqzN+/Pgcdthh1T00AJaiTz/9NG+99Vbl5++8805efPHFNGnSJK1bt67GkcGyy3Td8BNzxRVX5IILLsjEiROz3nrr5aKLLspWW21V3cMCYCl69NFHs80228yzvFevXhk8ePDSHxD8CAgjAACg8LzHCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwhNGAABA4QkjAACg8IQRAABQeMIIgB+9M844IxtttFHl5717907Pnj2X+jjGjh2bsrKyvPjii0v92AD8MMIIgCWmd+/eKSsrS1lZWcrLy9OmTZuccMIJmT59+hI97iWXXJLBgwcv1LpiBoAkqVXdAwDgp23HHXfMddddl1mzZuWJJ57IQQcdlOnTp+fKK6+sst6sWbNSXl6+WI7ZuHHjxbIfAIrDHSMAlqiKioq0bNkyq6yySvbee+/ss88+GTp0aOXjb4MGDUqbNm1SUVGRUqmUjz76KIccckiaN2+eRo0aZdttt83IkSOr7PO8885LixYt0rBhwxx44IH5/PPPq7z+zUfp5syZk/PPPz9rrrlmKioq0rp165xzzjlJktVXXz1J0r59+5SVlWXrrbeu3O66667Luuuumzp16mSdddbJFVdcUeU4zz77bNq3b586depk4403zgsvvLAYrxwAS5M7RgAsVXXr1s2sWbOSJG+99VZuvfXW3HHHHalZs2aSZKeddkqTJk1y3333pXHjxrnqqquy3XbbZfTo0WnSpEluvfXWnH766bn88svTuXPnDBkyJJdeemnatGmzwGOecsopueaaa3LRRRdlyy23zMSJE/P6668n+SpuNt100zz44INp165dateunSS55pprcvrpp+eyyy5L+/bt88ILL+Tggw9O/fr106tXr0yfPj2//OUvs+222+avf/1r3nnnnRx99NFL+OoBsKQIIwCWmmeffTZ/+9vfst122yVJvvjiiwwZMiTNmjVLkjz88MN5+eWXM3ny5FRUVCRJLrzwwgwdOjS33357DjnkkFx88cXp06dPDjrooCTJ2WefnQcffHCeu0ZzffLJJ7nkkkty2WWXpVevXkmSNdZYI1tuuWWSVB67adOmadmyZeV2Z511Vv785z/nV7/6VZKv7iy99tprueqqq9KrV6/ceOONmT17dgYNGpR69eqlXbt2ee+993L44Ycv7ssGwFLgUToAlqh77rknDRo0SJ06ddKpU6dstdVW6d+/f5Jk1VVXrQyTJHn++efz6aefpmnTpmnQoEHlxzvvvJO33347STJq1Kh06tSpyjG++fnXjRo1KjNnzqyMsYXxwQcf5N13382BBx5YZRxnn312lXFsuOGGqVev3kKNA4BlmztGACxR22yzTa688sqUl5dnpZVWqjLBQv369ausO2fOnKy44op59NFH59nPcsst972OX7du3UXeZs6cOUm+epxus802q/La3Ef+SqXS9xoPAMsmYQTAElW/fv2sueaaC7Vuhw4dMmnSpNSqVSurrbbafNdZd9118/TTT2f//fevXPb0008vcJ8/+9nPUrdu3Tz00EOVj9993dz3FM2ePbtyWYsWLdKqVauMGTMm++yzz3z327Zt2wwZMiQzZsyojK9vGwcAyzaP0gGwzNh+++3TqVOn9OzZM//85z8zduzYPPnkk/m///u/PPfcc0mSo48+OoMGDcqgQYMyevTonH766Xn11VcXuM86derk5JNPzkknnZQbbrghb7/9dp5++ukMHDgwSdK8efPUrVs3w4cPz/vvv5+PPvooyVe/NLZfv3655JJLMnr06Lz88su57rrr8pe//CVJsvfee6dGjRo58MAD89prr+W+++7LhRdeuISvEABLijACYJlRVlaW++67L1tttVX69OmTtdZaK3vttVfGjh2bFi1aJEn23HPP9O3bNyeffHI6duyYcePGfeeEB6eddlqOP/749O3bN+uuu2723HPPTJ48OUlSq1atXHrppbnqqquy0korZZdddkmSHHTQQbn22mszePDgrL/++unSpUsGDx5cOb13gwYNcvfdd+e1115L+/btc+qpp+b8889fglcHgCWprOQhaQAAoODcMQIAAApPGAEAAIUnjAAAgMITRgAAQOEJIwAAoPCEEQAAUHjCCAAAKDxhBAAAFJ4wAgAACk8YAQAAhSeMAACAwvv/AAunxC0cd+MxAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 1000x700 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC AUC Score (Adversarial Data): 0.9999858430228603\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x700 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import pickle\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from sklearn.metrics import classification_report, confusion_matrix, roc_auc_score, roc_curve\n",
    "\n",
    "\n",
    "# Load the adversarial datasets\n",
    "adv_test = pd.read_csv(\"data/pgd_test.csv\")\n",
    "\n",
    "x_test_adv = adv_test.iloc[:, :-1]  # Features\n",
    "y_test_adv = adv_test.iloc[:, -1]   # Labels\n",
    "\n",
    "# Ensure labels are properly formatted\n",
    "y_test_adv_np = y_test_adv.squeeze().values\n",
    "\n",
    "# Load the adversarial XGBoost model from .pkl file\n",
    "with open('best_xgboost_adversarial_model.pkl', 'rb') as f:\n",
    "    model_adv = pickle.load(f)\n",
    "\n",
    "# Make predictions\n",
    "y_pred_prob_adv = model_adv.predict_proba(x_test_adv)[:, 1]\n",
    "y_pred_adv = (y_pred_prob_adv > 0.5).astype(int)\n",
    "\n",
    "# Compute metrics\n",
    "print(\"Classification Report (Adversarial Data):\")\n",
    "print(classification_report(y_test_adv_np, y_pred_adv))\n",
    "\n",
    "print(\"Confusion Matrix (Adversarial Data):\")\n",
    "cm_adv = confusion_matrix(y_test_adv_np, y_pred_adv)\n",
    "print(cm_adv)\n",
    "\n",
    "# Plot Confusion Matrix (Adversarial Data)\n",
    "plt.figure(figsize=(10, 7))\n",
    "sns.heatmap(cm_adv, annot=True, fmt='d', cmap='Blues', cbar=False)\n",
    "plt.xlabel('Predicted')\n",
    "plt.ylabel('Actual')\n",
    "plt.title('Confusion Matrix (Adversarial Data)')\n",
    "plt.show()\n",
    "\n",
    "# ROC AUC (Adversarial Data)\n",
    "roc_auc_adv = roc_auc_score(y_test_adv_np, y_pred_prob_adv)\n",
    "print(f'ROC AUC Score (Adversarial Data): {roc_auc_adv}')\n",
    "\n",
    "# ROC Curve (Adversarial Data)\n",
    "fpr_adv, tpr_adv, _ = roc_curve(y_test_adv_np, y_pred_prob_adv)\n",
    "plt.figure(figsize=(10, 7))\n",
    "plt.plot(fpr_adv, tpr_adv, marker='.')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('ROC Curve (Adversarial Data)')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.11.0 ('base')",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.0"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "40d3a090f54c6569ab1632332b64b2c03c39dcf918b08424e98f38b5ae0af88f"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}