main.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "cac7d0d6-851a-41c9-9440-16c39887dd70",
   "metadata": {},
   "source": [
    "## Kaggle NLP competition\n",
    "https://www.kaggle.com/competitions/nlp-getting-started\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "63728142-e643-493f-806a-c6d6d4d11d9a",
   "metadata": {
    "tags": []
   },
   "source": [
    "<a id='TOC'/>\n",
    "\n",
    "### TOC (table of content)\n",
    "1. [Loading packages](#loading_packages)\n",
    "2. [Loading data](#loading_data)\n",
    "3. [Vectorizer](#vectorizer)\n",
    "4. [Model](#model)\n",
    "5. [Improvements](#improvements)\n",
    "6. [Neural network](#neural_network)\n",
    "7. [Decision tree classifier](#decision_tree)\n",
    "8. [Random forest](#random_forest)\n",
    "9. [KNN classifier](#knn)\n",
    "10. [AdaBoost classifier](#adaboo)\n",
    "11. [Stacking](#stacking)\n",
    "12. [Voting](#voting)\n",
    "13. [Conclusion](#conclusion)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d3864f76-0668-48e0-960e-b6d864c1ae39",
   "metadata": {},
   "source": [
    "<a id='loading_packages'/>\n",
    "\n",
    "### Loading packages\n",
    "[Back to TOC](#TOC)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "id": "55e03f18-ce11-40c6-bc01-b0bb230b4d6d",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package wordnet to /home/karol/nltk_data...\n",
      "[nltk_data]   Package wordnet is already up-to-date!\n",
      "[nltk_data] Downloading package omw-1.4 to /home/karol/nltk_data...\n",
      "[nltk_data]   Package omw-1.4 is already up-to-date!\n"
     ]
    }
   ],
   "source": [
    "import numpy as np # linear algebra\n",
    "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n",
    "\n",
    "\n",
    "from sklearn.model_selection import train_test_split, GridSearchCV\n",
    "from sklearn.feature_extraction.text import CountVectorizer, TfidfVectorizer\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.metrics import roc_auc_score, accuracy_score, confusion_matrix, ConfusionMatrixDisplay\n",
    "from sklearn.ensemble import RandomForestClassifier, StackingClassifier, AdaBoostClassifier, VotingClassifier\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "\n",
    "import nltk\n",
    "nltk.download('wordnet')\n",
    "nltk.download('omw-1.4')\n",
    "\n",
    "\n",
    "# Deep learning: \n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Embedding, GlobalAveragePooling1D, GlobalMaxPooling1D, Dropout\n",
    "from keras.callbacks import EarlyStopping, ModelCheckpoint, ReduceLROnPlateau\n",
    "import tensorflow as tf\n",
    "\n",
    "from tensorflow.keras.layers.experimental.preprocessing import TextVectorization\n",
    "\n",
    "import io\n",
    "import re\n",
    "import string\n",
    "\n",
    "\n",
    "\n",
    "from sklearn.tree import DecisionTreeClassifier, plot_tree"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8c87d160-5b4b-453b-8c29-bf5c9029ca15",
   "metadata": {},
   "source": [
    "<a id='loading_data'/>\n",
    "\n",
    "### Loading data\n",
    "[Back to TOC](#TOC)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "f740b176-5bbf-48bc-bfa8-ab6664258fd1",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Set Shape = (7613, 5)\n",
      "Training Set Memory Usage = 0.29 MB\n"
     ]
    }
   ],
   "source": [
    "df_train = pd.read_csv(\"train.csv\")\n",
    "df_test = pd.read_csv(\"test.csv\")\n",
    "df=df_train.copy()\n",
    "print('Training Set Shape = {}'.format(df_train.shape))\n",
    "print('Training Set Memory Usage = {:.2f} MB'.format(df_train.memory_usage().sum() / 1024**2))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f4738fe9-37c9-4584-b5ac-e483611892e3",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Let's have a look at the train and test dataset.\n",
    "\n",
    "#### They contain:\n",
    "- id\n",
    "- keyword: A keyword from that tweet (although this may be blank!)\n",
    "- location: The location the tweet was sent from (may also be blank)\n",
    "- text: The text of a tweet\n",
    "- target: 1 if the tweet is a real disaster or 0 if not"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "7a347d79-6ff3-4ff3-bd76-bbc647dd20d5",
   "metadata": {
    "tags": []
   },
   "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>id</th>\n",
       "      <th>keyword</th>\n",
       "      <th>location</th>\n",
       "      <th>text</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Our Deeds are the Reason of this #earthquake M...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Forest fire near La Ronge Sask. Canada</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>All residents asked to 'shelter in place' are ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>13,000 people receive #wildfires evacuation or...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Just got sent this photo from Ruby #Alaska as ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   id keyword location                                               text  \\\n",
       "0   1     NaN      NaN  Our Deeds are the Reason of this #earthquake M...   \n",
       "1   4     NaN      NaN             Forest fire near La Ronge Sask. Canada   \n",
       "2   5     NaN      NaN  All residents asked to 'shelter in place' are ...   \n",
       "3   6     NaN      NaN  13,000 people receive #wildfires evacuation or...   \n",
       "4   7     NaN      NaN  Just got sent this photo from Ruby #Alaska as ...   \n",
       "\n",
       "   target  \n",
       "0       1  \n",
       "1       1  \n",
       "2       1  \n",
       "3       1  \n",
       "4       1  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "ccf6c4e8-a1af-4af6-a80d-320dc033df4c",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The data contains 3271 actual disasters (42.97% of all data).\n"
     ]
    }
   ],
   "source": [
    "print(\n",
    "    \"The data contains {} actual disasters ({:.2%} of all data).\".format(\n",
    "        df[df.target == 1].shape[0],\n",
    "        df[df.target == 1].shape[0] / df.shape[0],\n",
    "    )\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "8eb36cd9-8889-4759-a687-5a1778a3d2b9",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Examples of actual disaster tweet: \n",
      "    Refugio oil spill may have been costlier bigger than projected http://t.co/xcoLwUGFjg\n",
      "    Investigators say a fatal Virgin Galactic spaceship crash last year was caused by structural failure after the coÛ_\n",
      "\n",
      "Examples of a tweet with no disaster behind it: \n",
      "    wreck? wreck wreck wreck wreck wreck wreck wreck wreck wreck wreck wreck wreck?\n",
      "    I liked a @YouTube video http://t.co/5fR41TPzte Thorin's Thoughts - Riot and Sandbox Mode (LoL)\n"
     ]
    }
   ],
   "source": [
    "print(\n",
    "    \"Examples of actual disaster tweet: \\n    {}\\n    {}\".format(\n",
    "        df[df.target == 1].sample(1).text.iloc[0],\n",
    "        df[df.target == 1].sample(1).text.iloc[0],\n",
    "    )\n",
    ")\n",
    "print(\n",
    "    \"\\nExamples of a tweet with no disaster behind it: \\n    {}\\n    {}\".format(\n",
    "        df[df.target == 0].sample(1).text.iloc[0],\n",
    "        df[df.target == 0].sample(1).text.iloc[0],\n",
    "    )\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "83ec6860-de4c-4ee3-9e6c-bb8a1e78bd31",
   "metadata": {},
   "source": [
    "<a id='vectorizer'/>\n",
    "\n",
    "### Vectorizer\n",
    "[Back to TOC](#TOC)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "id": "723ff8da-6fc3-489f-8dea-1e7661dbc79f",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(df_train[\"text\"], df_train[\"target\"], random_state=777,test_size=0.2, stratify=df_train[\"target\"])\n",
    "\n",
    "\n",
    "\n",
    "X_train, X_val, y_train, y_val   = train_test_split(X_train, y_train, random_state=777, test_size=0.2, stratify=y_train)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "id": "48b4034d-3211-4c9a-bc20-7bebb67ff20a",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Length of train set: 4872\n",
      " Length of test set: 1218\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\n",
    "    \"Length of train set: {}\\n Length of test set: {}\\n\".format(len(X_train), len(X_val))\n",
    ")\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ae439de9-c764-4abe-a18b-33643d9a1a26",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Transform the text so that each word is a separate feature and count how many times that word occurs. (Save as a sparse matrix.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "id": "0a3af641-3402-443d-9cec-ee3fd9788057",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-40 {color: black;background-color: white;}#sk-container-id-40 pre{padding: 0;}#sk-container-id-40 div.sk-toggleable {background-color: white;}#sk-container-id-40 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-40 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-40 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-40 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-40 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-40 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-40 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-40 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-40 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-40 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-40 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-40 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-40 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-40 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-40 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-40 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-40 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-40 div.sk-item {position: relative;z-index: 1;}#sk-container-id-40 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-40 div.sk-item::before, #sk-container-id-40 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-40 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-40 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-40 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-40 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-40 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-40 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-40 div.sk-label-container {text-align: center;}#sk-container-id-40 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-40 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-40\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>CountVectorizer()</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-65\" type=\"checkbox\" checked><label for=\"sk-estimator-id-65\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">CountVectorizer</label><div class=\"sk-toggleable__content\"><pre>CountVectorizer()</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "CountVectorizer()"
      ]
     },
     "execution_count": 175,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Fit the CountVectorizer to the training data\n",
    "vect = CountVectorizer().fit(X_train)\n",
    "\n",
    "# transform the text in the training data to a text-term matrix\n",
    "X_train_vectorized = vect.transform(X_train)\n",
    "\n",
    "vect"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8f999cf3-7a72-454f-8ba3-bbcabbf91f57",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Vocabulary sample:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "id": "a6c28941-ad3d-47fe-862f-4f5bed968025",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('delivers', 4000),\n",
       " ('dell', 4001),\n",
       " ('delmont', 4002),\n",
       " ('delta', 4003),\n",
       " ('deluded', 4004),\n",
       " ('deluge', 4005),\n",
       " ('deluged', 4006),\n",
       " ('delusion', 4007),\n",
       " ('deluxe', 4008),\n",
       " ('dem', 4009)]"
      ]
     },
     "execution_count": 176,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sorted(vect.vocabulary_.items())[4000:4010]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "id": "bb5aa2ab-a1fa-4d6a-a139-129662d0a520",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The training data contains 15759 different words (columns) across 4872 messages (rows).\n"
     ]
    }
   ],
   "source": [
    "full_vocab_len=len(vect.vocabulary_)\n",
    "\n",
    "print(\n",
    "    \"The training data contains {} different words (columns) across {} messages (rows).\".format( \n",
    "                                                            len(vect.vocabulary_.items()),\n",
    "                                                            X_train.shape[0]\n",
    "                                                           )\n",
    "     )"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aaac3a9a-1924-4ede-bf2a-ef160c05c176",
   "metadata": {},
   "source": [
    "### Show the (sparse) feature matrix:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "id": "ac6daa7d-66f8-4087-90f4-d477acd8c06e",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0, 0, 0, ..., 0, 0, 0],\n",
       "       [0, 0, 0, ..., 0, 0, 0],\n",
       "       [0, 0, 0, ..., 0, 0, 0],\n",
       "       ...,\n",
       "       [0, 0, 0, ..., 0, 0, 0],\n",
       "       [0, 0, 0, ..., 0, 0, 0],\n",
       "       [0, 0, 0, ..., 0, 0, 0]])"
      ]
     },
     "execution_count": 178,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_vectorized.toarray()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4d4d3fb0-fb3a-40ac-9d34-d6e2d308339b",
   "metadata": {
    "tags": []
   },
   "source": [
    "<a id='model'/>\n",
    "\n",
    "### Model\n",
    "[Back to TOC](#TOC)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3cd53752-63cb-4c88-a81b-0deba9510448",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Let's train a logistic regression model with the train data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "id": "4b006e1d-83a0-46d4-9870-73af64732bad",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy = 0.79317\n",
      "AUC = 0.837\n"
     ]
    }
   ],
   "source": [
    "log_reg_model = LogisticRegression(max_iter=777)\n",
    "log_reg_model.fit(X_train_vectorized, y_train)\n",
    "\n",
    "# Predict the transformed validation texts\n",
    "predictions = log_reg_model.predict(vect.transform(X_test))\n",
    "predict_probab = log_reg_model.predict_proba(vect.transform(X_test))[:,1]\n",
    "\n",
    "print(\"Accuracy = {:.5f}\".format(accuracy_score(y_test, predictions)))\n",
    "print(\"AUC = {:.3f}\".format(roc_auc_score(y_test, predict_probab)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "id": "0cac7d29-8a2c-4c6f-b954-29fed2147cb2",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7fdb3921cf90>"
      ]
     },
     "execution_count": 180,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "confusion_matrix_model = ConfusionMatrixDisplay(confusion_matrix(y_test, predictions))\n",
    "confusion_matrix_model.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ba548fd2-40dc-4d33-8165-d58633ccf0ce",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Let's check which words are most probably associated with actual disasters and which are not."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "id": "acc4ff45-119d-4358-b583-dfb39b7b6092",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Words with the smallest Coefficients (i.e. not indicative of a real disaster):\n",
      "['full' 'song' 'new' 'panic' 'ebay' 'sure' 'let' 'ass' 'obliteration'\n",
      " 'bags' 'poll' 'blight' 'collide' 'cake' 'ruin' 'flattened' 'head' 'lmao'\n",
      " 'nowplaying' 'upheaval']\n",
      "\n",
      "Words with the largest Coefficients (i.e. very indicative of a real disaster): \n",
      "['hiroshima' 'storm' 'earthquake' 'fires' 'wildfire' 'california' 'shot'\n",
      " 'train' 'typhoon' 'volcano' 'floods' 'drought' 'bombing' 'hurricane'\n",
      " 'evacuated' 'spill' 'casualties' 'forest' 'suicide' 'plane']\n"
     ]
    }
   ],
   "source": [
    "# list of words (features)\n",
    "feature_names = np.array(vect.get_feature_names_out())\n",
    "\n",
    "#list of corresponding coefficients\n",
    "sorted_coef_index = log_reg_model.coef_[0].argsort()\n",
    "\n",
    "print(\"Words with the smallest Coefficients (i.e. not indicative of a real disaster):\\n{}\\n\".format(feature_names[sorted_coef_index[:20]]))\n",
    "print(\"Words with the largest Coefficients (i.e. very indicative of a real disaster): \\n{}\".format(feature_names[sorted_coef_index[:-21:-1]]))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0680b14f-dd4c-4865-b9ec-4351f0e729da",
   "metadata": {},
   "source": [
    "<a id='improvements'/>\n",
    "\n",
    "### Improvements\n",
    "[Back to TOC](#TOC)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "983d4f75-8800-4673-849b-305ed86303de",
   "metadata": {},
   "source": [
    "### Let's try to improve our prediction by considering term frequency using TF-IDF "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 184,
   "id": "c60ccf56-e382-42a2-bea1-64422435c470",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('age', 150),\n",
       " ('ago', 151),\n",
       " ('agree', 152),\n",
       " ('agreed', 153),\n",
       " ('ah', 154),\n",
       " ('ahead', 155),\n",
       " ('aid', 156),\n",
       " ('aim', 157),\n",
       " ('ain', 158),\n",
       " ('air', 159)]"
      ]
     },
     "execution_count": 184,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Consider only words that occur in at least 3 tweets.\n",
    "vect = TfidfVectorizer(min_df=3).fit(X_train)\n",
    "\n",
    "\n",
    "X_train_vectorized = vect.transform(X_train)\n",
    "\n",
    "# let's look of some of the words gathered with this method\n",
    "sorted(vect.vocabulary_.items())[150:160]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "id": "30ba2282-88ea-47e2-b67a-47dc240c87a6",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The training data now contains 3214 different words (columns) across 4872 messages (rows).\n"
     ]
    }
   ],
   "source": [
    "print(\n",
    "    \"The training data now contains {} different words (columns) across {} messages (rows).\".format( \n",
    "                                                            len(vect.vocabulary_.items()),\n",
    "                                                            X_train.shape[0]\n",
    "                                                           )\n",
    "     )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "id": "d586651a-d3f1-4136-bd07-54e8ade35ee8",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Smallest tfidf:\n",
      "['ksawlyux02' 'sh' 'utc2015' 'time2015' 'sm' 'destiny' 'enemy' 'blessings'\n",
      " 'curved' '300w' '53inch' 'offroad' 'arwx' 'eq' 'disea' 'realise'\n",
      " 'prebreak' '18w' '2pcs' 'handbag']\n",
      "\n",
      "Largest tfidf: \n",
      "['hurricane' 'on' 'mudslide' 'girl' 'sinking' 'meltdown' 'twister'\n",
      " 'flattened' 'flames' 'no' 'emergency' 'fatality' 'traumatised' 'wreck'\n",
      " 'trauma' 'weapon' 'weapons' 'survive' 'obliterated' 'thanks']\n"
     ]
    }
   ],
   "source": [
    "# list of words (features)\n",
    "feature_names = np.array(vect.get_feature_names_out())\n",
    "\n",
    "#list of corresponding tfidf values\n",
    "sorted_tfidf_index = X_train_vectorized.toarray().max(0).argsort()\n",
    "\n",
    "# print words with highest and lowest tfidf values\n",
    "print(\"Smallest tfidf:\\n{}\\n\".format(feature_names[sorted_tfidf_index[:20]]))\n",
    "print(\"Largest tfidf: \\n{}\".format(feature_names[sorted_tfidf_index[:-21:-1]]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "id": "92ea0875-e75b-4f71-8342-2e67e041f458",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy = 0.79054\n",
      "AUC = 0.841\n"
     ]
    }
   ],
   "source": [
    "log_reg_model_TF_IDF = LogisticRegression(max_iter=777)\n",
    "log_reg_model_TF_IDF.fit(X_train_vectorized, y_train)\n",
    "\n",
    "# Predict the transformed validation texts\n",
    "predictions = log_reg_model_TF_IDF.predict(vect.transform(X_test))\n",
    "predict_probab = log_reg_model_TF_IDF.predict_proba(vect.transform(X_test))[:,1]\n",
    "\n",
    "print(\"Accuracy = {:.5f}\".format(accuracy_score(y_test, predictions)))\n",
    "print(\"AUC = {:.3f}\".format(roc_auc_score(y_test, predict_probab)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "id": "e22db5a1-ef49-40d5-95ad-134665b724b2",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7fdb38f62490>"
      ]
     },
     "execution_count": 188,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "confusion_matrix_model = ConfusionMatrixDisplay(confusion_matrix(y_test, predictions))\n",
    "confusion_matrix_model.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "id": "9b5b42d0-30a7-4a2f-ab44-12e068245556",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Words with the smallest Coefficients (i.e. not indicative of a real disaster):\n",
      "['you' 'new' 'full' 'my' 'let' 'youtube' 'body' 'panic' 'or' 'bags' 'he'\n",
      " 'ebay' 'blight' 'song' 'obliteration' 'ruin' 'flattened' 'collide'\n",
      " 'demolish' 'me']\n",
      "\n",
      "Words with the largest Coefficients (i.e. very indicative of a real disaster): \n",
      "['hiroshima' 'in' 'california' 'fires' 'storm' 'train' 'killed' 'wildfire'\n",
      " 'bombing' 'earthquake' 'fire' 'suicide' 'buildings' 'typhoon' 'disaster'\n",
      " 'forest' 'japan' 'evacuated' 'car' 'accident']\n"
     ]
    }
   ],
   "source": [
    "# list of words (features)\n",
    "feature_names = np.array(vect.get_feature_names_out())\n",
    "\n",
    "#list of corresponding coefficients\n",
    "sorted_coef_index = log_reg_model_TF_IDF.coef_[0].argsort()\n",
    "\n",
    "print(\"Words with the smallest Coefficients (i.e. not indicative of a real disaster):\\n{}\\n\".format(feature_names[sorted_coef_index[:20]]))\n",
    "print(\"Words with the largest Coefficients (i.e. very indicative of a real disaster): \\n{}\".format(feature_names[sorted_coef_index[:-21:-1]]))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "14658900-3888-4a89-9012-82b0169e6d07",
   "metadata": {},
   "source": [
    "### Using TF-IDF, we have increased our AUC score slightly. But we have lost a few true negatives."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "850641f2-2910-4c52-b2a6-f48d8cde70af",
   "metadata": {},
   "source": [
    "### Let's make the texts more comparable by using stemming."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "id": "854d6a55-0554-47ff-aa6d-60f72697f6be",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "stemmer = nltk.PorterStemmer()\n",
    "cv_analyzer = TfidfVectorizer(min_df=2).build_analyzer()\n",
    "\n",
    "def stemmed_words(doc):\n",
    "    return (stemmer.stem(w) for w in cv_analyzer(doc))\n",
    "\n",
    "stem_vectorizer = TfidfVectorizer(min_df=2, analyzer = stemmed_words)\n",
    "X_train_stem_vectorized = stem_vectorizer.fit_transform(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 193,
   "id": "4de055d9-123f-487b-a045-1d3475d67beb",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy = 0.78858\n",
      "AUC = 0.847\n"
     ]
    }
   ],
   "source": [
    "log_reg_model_stem = LogisticRegression(max_iter=777)\n",
    "log_reg_model_stem.fit(X_train_stem_vectorized, y_train)\n",
    "\n",
    "# Predict the transformed validation texts\n",
    "predictions = log_reg_model_stem.predict(stem_vectorizer.transform(X_test))\n",
    "predict_probab = log_reg_model_stem.predict_proba(stem_vectorizer.transform(X_test))[:,1]\n",
    "\n",
    "print(\"Accuracy = {:.5f}\".format(accuracy_score(y_test, predictions)))\n",
    "print(\"AUC = {:.3f}\".format(roc_auc_score(y_test, predict_probab)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 194,
   "id": "2a8da7a5-45fa-4c53-963e-73e48510bdd7",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7fdaa065c1d0>"
      ]
     },
     "execution_count": 194,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "confusion_matrix_model = ConfusionMatrixDisplay(confusion_matrix(y_test, predictions))\n",
    "confusion_matrix_model.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "75fdcf64-2ab4-478f-9978-06d9cbe4631c",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Stemming further increases the AUC. But now we have lost a few true positives. What about Lemmatization?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "id": "518686d6-453d-4b85-808f-363d3e25378e",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "WNlemma = nltk.WordNetLemmatizer()\n",
    "cv_analyzer = TfidfVectorizer(min_df=4).build_analyzer()\n",
    "\n",
    "\n",
    "def lemmatize_word(doc):\n",
    "    return (WNlemma.lemmatize(t) for t in cv_analyzer(doc))\n",
    "\n",
    "lemm_vectorizer = TfidfVectorizer(min_df=4, analyzer = lemmatize_word)\n",
    "\n",
    "\n",
    "X_train_lemm_vectorized = lemm_vectorizer.fit_transform(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "id": "aad90391-dbb5-4992-b0d0-a9212dbd1fd4",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy = 0.78267\n",
      "AUC = 0.843\n"
     ]
    }
   ],
   "source": [
    "log_reg_model_lemm = LogisticRegression(max_iter=777)\n",
    "log_reg_model_lemm.fit(X_train_lemm_vectorized, y_train)\n",
    "\n",
    "# Predict the transformed validation texts\n",
    "predictions = log_reg_model_lemm.predict(lemm_vectorizer.transform(X_test))\n",
    "predict_probab = log_reg_model_lemm.predict_proba(lemm_vectorizer.transform(X_test))[:,1]\n",
    "\n",
    "print(\"Accuracy = {:.5f}\".format(accuracy_score(y_test, predictions)))\n",
    "print(\"AUC = {:.3f}\".format(roc_auc_score(y_test, predict_probab)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 197,
   "id": "84b763bb-b006-4784-b450-5f135cf0a840",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7fdaa069b7d0>"
      ]
     },
     "execution_count": 197,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "confusion_matrix_model = ConfusionMatrixDisplay(confusion_matrix(y_test, predictions))\n",
    "confusion_matrix_model.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "62ba160c-c6eb-4b86-90b3-8c2a6bc2d076",
   "metadata": {},
   "source": [
    "### Nope. Lemmatization yields the worst results yet."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "55f83608-1d8b-40c9-ac23-8af8585f7f32",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Next, let's test a neural network."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "241ff71e-e4df-4e4d-91a7-48f4d88a191e",
   "metadata": {},
   "source": [
    "<a id='neural_network'/>\n",
    "\n",
    "### Neural network\n",
    "[Back to TOC](#TOC)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "4593aa93-a6f4-4c99-a616-e050f686b033",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train size: 4872\n",
      "test size: 1523\n",
      "val size: 1218\n"
     ]
    }
   ],
   "source": [
    "\n",
    "print('train size:', len(y_train))\n",
    "\n",
    "print('test size:', len(y_test))\n",
    "\n",
    "print('val size:', len(y_val))\n",
    "\n",
    "\n",
    "# convert datasets for use in Keras\n",
    "\n",
    "\n",
    "dataset_train = tf.data.Dataset.from_tensor_slices((X_train, y_train))\n",
    "dataset_val = tf.data.Dataset.from_tensor_slices((X_val, y_val))\n",
    "dataset_test = tf.data.Dataset.from_tensor_slices((X_test, y_test))\n",
    "\n",
    "dataframe_train=pd.concat([X_train,y_train], axis=1)\n",
    "dataframe_test=pd.concat([X_test,y_test], axis=1)\n",
    "\n",
    "# create batches of the datasets for training, validation and testing\n",
    "\n",
    "batch_size=500\n",
    "dataset_train_batch = dataset_train.batch(batch_size=batch_size)\n",
    "dataset_val_batch = dataset_val.batch(batch_size=batch_size)\n",
    "dataset_test_batch = dataset_test.batch(batch_size=batch_size)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d297aa4d-0bd6-4660-8a47-fa175fb3c82b",
   "metadata": {
    "tags": []
   },
   "source": [
    "Define Vecorizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "46eb8373-b42a-443c-ada6-b659bc7a1b5c",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def custom_standardization(input_data):\n",
    "    \"\"\"Text preprocessing: lowercases, no punctuation\n",
    "\n",
    "    Args:\n",
    "        input_data (tf.dataframe): [text, formated as tf.string]\n",
    "\n",
    "    Returns:\n",
    "        [tf.dataframe]: [preprocessed text]\n",
    "    \"\"\"\n",
    "    text_lower = tf.strings.lower(input_data)\n",
    "    return tf.strings.regex_replace(text_lower, '[%s]' % re.escape(string.punctuation), '')\n",
    "\n",
    "# get the number of words in each message\n",
    "sequence_len = int(dataframe_train.text.apply(lambda x: len(x.split())).max()) \n",
    "\n",
    "# define the text vectorization layer \n",
    "vectorize_layer = TextVectorization(\n",
    "    standardize=custom_standardization,\n",
    "    max_tokens=full_vocab_len,\n",
    "    output_mode='int',\n",
    "    output_sequence_length=sequence_len)\n",
    "\n",
    "# Make a text-only dataset (without labels)\n",
    "train_text = dataset_train.map(lambda x, y: x)\n",
    "vectorize_layer.adapt(train_text)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "85186155-02a7-454f-8a63-d741a09265a7",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1138 --->  terror\n",
      " 327 --->  survivors\n",
      "Vocabulary size: 15759\n"
     ]
    }
   ],
   "source": [
    "print(\"1138 ---> \",vectorize_layer.get_vocabulary()[1138])\n",
    "print(\" 327 ---> \",vectorize_layer.get_vocabulary()[327])\n",
    "print('Vocabulary size: {}'.format(len(vectorize_layer.get_vocabulary())))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b4bdf2d5-02ab-4aa4-9c93-17ddaaad016f",
   "metadata": {},
   "source": [
    "### Define the model. Note that it uses word embeddings."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "d5c3270a-d278-4df5-8f4c-3966267cde8a",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "tf.keras.backend.clear_session()\n",
    "\n",
    "embedding_dim=3\n",
    "model_nn = Sequential([\n",
    "  vectorize_layer,\n",
    "  Embedding(full_vocab_len, embedding_dim, name=\"embedding\"),\n",
    "  GlobalAveragePooling1D(),\n",
    "  Dropout(0.5),\n",
    "  Dense(500, activation='relu'),\n",
    "  Dropout(0.25),\n",
    "#  Dense(16, activation='elu'),\n",
    "  Dense(1, activation='sigmoid')\n",
    "])\n",
    "\n",
    "model_nn.compile(optimizer='adam',\n",
    "              loss=tf.keras.losses.BinaryCrossentropy(),\n",
    "              metrics=['AUC'])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "492c52bc-b6ad-4890-9142-15f5f0362c9a",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "10/10 [==============================] - 2s 48ms/step - loss: 0.6887 - auc: 0.5035 - val_loss: 0.6843 - val_auc: 0.5009\n",
      "Epoch 2/100\n",
      "10/10 [==============================] - 0s 23ms/step - loss: 0.6833 - auc: 0.5093 - val_loss: 0.6827 - val_auc: 0.5768\n",
      "Epoch 3/100\n",
      "10/10 [==============================] - 0s 31ms/step - loss: 0.6828 - auc: 0.5387 - val_loss: 0.6820 - val_auc: 0.6151\n",
      "Epoch 4/100\n",
      "10/10 [==============================] - 0s 28ms/step - loss: 0.6813 - auc: 0.5772 - val_loss: 0.6806 - val_auc: 0.6727\n",
      "Epoch 5/100\n",
      "10/10 [==============================] - 0s 28ms/step - loss: 0.6797 - auc: 0.6060 - val_loss: 0.6785 - val_auc: 0.7185\n",
      "Epoch 6/100\n",
      "10/10 [==============================] - 0s 19ms/step - loss: 0.6758 - auc: 0.6625 - val_loss: 0.6752 - val_auc: 0.7524\n",
      "Epoch 7/100\n",
      "10/10 [==============================] - 0s 18ms/step - loss: 0.6723 - auc: 0.6748 - val_loss: 0.6701 - val_auc: 0.7806\n",
      "Epoch 8/100\n",
      "10/10 [==============================] - 0s 29ms/step - loss: 0.6649 - auc: 0.7219 - val_loss: 0.6621 - val_auc: 0.8074\n",
      "Epoch 9/100\n",
      "10/10 [==============================] - 0s 35ms/step - loss: 0.6522 - auc: 0.7665 - val_loss: 0.6502 - val_auc: 0.8212\n",
      "Epoch 10/100\n",
      "10/10 [==============================] - 0s 34ms/step - loss: 0.6367 - auc: 0.8012 - val_loss: 0.6347 - val_auc: 0.8297\n",
      "Epoch 11/100\n",
      "10/10 [==============================] - 0s 26ms/step - loss: 0.6152 - auc: 0.8380 - val_loss: 0.6162 - val_auc: 0.8343\n",
      "Epoch 12/100\n",
      "10/10 [==============================] - 0s 28ms/step - loss: 0.5873 - auc: 0.8707 - val_loss: 0.5949 - val_auc: 0.8368\n",
      "Epoch 13/100\n",
      "10/10 [==============================] - 0s 24ms/step - loss: 0.5613 - auc: 0.8774 - val_loss: 0.5722 - val_auc: 0.8404\n",
      "Epoch 14/100\n",
      "10/10 [==============================] - 0s 23ms/step - loss: 0.5237 - auc: 0.9010 - val_loss: 0.5501 - val_auc: 0.8435\n",
      "Epoch 15/100\n",
      "10/10 [==============================] - 0s 21ms/step - loss: 0.4943 - auc: 0.9037 - val_loss: 0.5289 - val_auc: 0.8473\n",
      "Epoch 16/100\n",
      "10/10 [==============================] - 0s 28ms/step - loss: 0.4592 - auc: 0.9132 - val_loss: 0.5130 - val_auc: 0.8505\n",
      "Epoch 17/100\n",
      "10/10 [==============================] - 0s 24ms/step - loss: 0.4243 - auc: 0.9261 - val_loss: 0.4946 - val_auc: 0.8550\n",
      "Epoch 18/100\n",
      "10/10 [==============================] - 0s 22ms/step - loss: 0.4001 - auc: 0.9273 - val_loss: 0.4871 - val_auc: 0.8582\n",
      "Epoch 19/100\n",
      "10/10 [==============================] - 0s 20ms/step - loss: 0.3827 - auc: 0.9305 - val_loss: 0.4756 - val_auc: 0.8610\n",
      "Epoch 20/100\n",
      "10/10 [==============================] - 0s 21ms/step - loss: 0.3561 - auc: 0.9406 - val_loss: 0.4685 - val_auc: 0.8632\n",
      "Epoch 21/100\n",
      "10/10 [==============================] - 0s 26ms/step - loss: 0.3335 - auc: 0.9475 - val_loss: 0.4604 - val_auc: 0.8644\n",
      "Epoch 22/100\n",
      "10/10 [==============================] - 0s 22ms/step - loss: 0.3202 - auc: 0.9514 - val_loss: 0.4554 - val_auc: 0.8660\n",
      "Epoch 23/100\n",
      "10/10 [==============================] - 0s 17ms/step - loss: 0.3022 - auc: 0.9569 - val_loss: 0.4561 - val_auc: 0.8672\n",
      "Epoch 24/100\n",
      "10/10 [==============================] - 0s 21ms/step - loss: 0.2920 - auc: 0.9589 - val_loss: 0.4485 - val_auc: 0.8682\n",
      "Epoch 25/100\n",
      "10/10 [==============================] - 0s 16ms/step - loss: 0.2769 - auc: 0.9626 - val_loss: 0.4563 - val_auc: 0.8687\n",
      "Epoch 26/100\n",
      "10/10 [==============================] - 0s 17ms/step - loss: 0.2648 - auc: 0.9644 - val_loss: 0.4498 - val_auc: 0.8690\n",
      "Epoch 27/100\n",
      "10/10 [==============================] - 0s 21ms/step - loss: 0.2475 - auc: 0.9707 - val_loss: 0.4510 - val_auc: 0.8687\n",
      "Epoch 28/100\n",
      "10/10 [==============================] - 0s 20ms/step - loss: 0.2412 - auc: 0.9706 - val_loss: 0.4544 - val_auc: 0.8687\n",
      "Epoch 29/100\n",
      "10/10 [==============================] - 0s 17ms/step - loss: 0.2365 - auc: 0.9709 - val_loss: 0.4527 - val_auc: 0.8681\n",
      "Epoch 30/100\n",
      "10/10 [==============================] - 0s 18ms/step - loss: 0.2213 - auc: 0.9763 - val_loss: 0.4543 - val_auc: 0.8679\n",
      "Epoch 31/100\n",
      "10/10 [==============================] - 0s 18ms/step - loss: 0.2140 - auc: 0.9768 - val_loss: 0.4590 - val_auc: 0.8676\n",
      "Epoch 32/100\n",
      "10/10 [==============================] - 0s 17ms/step - loss: 0.2110 - auc: 0.9773 - val_loss: 0.4595 - val_auc: 0.8667\n",
      "Epoch 33/100\n",
      "10/10 [==============================] - 0s 16ms/step - loss: 0.2018 - auc: 0.9788 - val_loss: 0.4654 - val_auc: 0.8657\n",
      "Epoch 34/100\n",
      "10/10 [==============================] - 0s 17ms/step - loss: 0.1963 - auc: 0.9800 - val_loss: 0.4716 - val_auc: 0.8654\n",
      "Epoch 35/100\n",
      "10/10 [==============================] - 0s 19ms/step - loss: 0.1881 - auc: 0.9818 - val_loss: 0.4697 - val_auc: 0.8649\n",
      "Epoch 36/100\n",
      "10/10 [==============================] - 0s 17ms/step - loss: 0.1815 - auc: 0.9822 - val_loss: 0.4759 - val_auc: 0.8645\n",
      "Epoch 37/100\n",
      "10/10 [==============================] - 0s 20ms/step - loss: 0.1737 - auc: 0.9845 - val_loss: 0.4849 - val_auc: 0.8636\n",
      "Epoch 38/100\n",
      "10/10 [==============================] - 0s 19ms/step - loss: 0.1730 - auc: 0.9839 - val_loss: 0.4927 - val_auc: 0.8633\n",
      "Epoch 39/100\n",
      "10/10 [==============================] - 0s 15ms/step - loss: 0.1712 - auc: 0.9833 - val_loss: 0.4890 - val_auc: 0.8635\n",
      "Epoch 40/100\n",
      "10/10 [==============================] - 0s 17ms/step - loss: 0.1692 - auc: 0.9849 - val_loss: 0.5027 - val_auc: 0.8626\n",
      "Epoch 41/100\n",
      "10/10 [==============================] - 0s 17ms/step - loss: 0.1662 - auc: 0.9839 - val_loss: 0.5036 - val_auc: 0.8625\n",
      "Epoch 42/100\n",
      "10/10 [==============================] - 0s 16ms/step - loss: 0.1530 - auc: 0.9875 - val_loss: 0.5082 - val_auc: 0.8622\n",
      "Epoch 43/100\n",
      "10/10 [==============================] - 0s 17ms/step - loss: 0.1528 - auc: 0.9862 - val_loss: 0.5142 - val_auc: 0.8615\n",
      "Epoch 44/100\n",
      "10/10 [==============================] - 0s 18ms/step - loss: 0.1540 - auc: 0.9863 - val_loss: 0.5273 - val_auc: 0.8606\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.src.callbacks.History at 0x7fdac2765750>"
      ]
     },
     "execution_count": 75,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# training the model\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "early_stopping_monitor = EarlyStopping(\n",
    "    monitor='val_loss',\n",
    "    min_delta=0,\n",
    "    patience=20,\n",
    "    verbose=0,\n",
    "    mode='auto',\n",
    "    baseline=None,\n",
    "    restore_best_weights=True\n",
    ")\n",
    "\n",
    "\n",
    "\n",
    "model_nn.fit(\n",
    "    dataset_train_batch,\n",
    "    validation_data=dataset_val_batch,\n",
    "    epochs=100,\n",
    "    callbacks=[early_stopping_monitor]\n",
    "    )\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "15f19c96-b40a-4a55-b5a4-e44d18c0c82e",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "fc51a7f4-898c-4c32-b820-4424b1970fa1",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3/3 - 0s - loss: 0.4485 - auc: 0.8682 - 28ms/epoch - 9ms/step\n"
     ]
    }
   ],
   "source": [
    "# calculating the loss and accuracy on the test set.\n",
    "loss, accuracy = model_nn.evaluate(dataset_val_batch, verbose=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "id": "ff9d9ee2-e486-4d46-b374-f6e947bbc5ac",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7fdac0309a90>"
      ]
     },
     "execution_count": 77,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfsAAAGwCAYAAACuFMx9AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAAA5I0lEQVR4nO3deXhU5fn/8c9kJ8tMCJLEQAggsqSyFSxMrQsaichlodBaLUJUtD8xQYWKaMuOgl+somgAFwRtpYgLtKIii4ILQSWKRcAoiyQsSVRMQgLZZs7vj5ixI6gZZpJh5rxfvc5V55znnLnTInfu+3nOORbDMAwBAICgFeLvAAAAQPMi2QMAEORI9gAABDmSPQAAQY5kDwBAkCPZAwAQ5Ej2AAAEuTB/B+ANp9Opw4cPKy4uThaLxd/hAAA8ZBiGjh07ppSUFIWENF/9WV1drdraWq+vExERoaioKB9E1LICOtkfPnxYqamp/g4DAOCloqIitW/fvlmuXV1drU5psSoudXh9reTkZO3fvz/gEn5AJ/u4uDhJ0oGPOsoay4wEgtPvuvb0dwhAs6lXnd7Va66/z5tDbW2tiksdOpDfUda4088VFcecSuv3pWpra0n2LamxdW+NDfHq/0DgTBZmCfd3CEDz+e6B7S0xFRsbZ1Fs3Ol/j1OenduxY0cdOHDgpP233nqrcnNzVV1drb/85S9asWKFampqlJmZqYULFyopKck1trCwUOPGjdNbb72l2NhYZWVlae7cuQoL8yx9B3SyBwCgqRyGUw4v3gbjMJwejf/www/lcHw/dfDpp5/q8ssv1x/+8AdJ0oQJE/Tqq6/qhRdekM1mU05OjkaMGKH33nuv4fscDg0dOlTJycnasmWLjhw5ojFjxig8PFxz5szxKBaSPQDAFJwy5NTpZ3tPz23btq3b5/vvv1/nnHOOLr74YpWXl2vJkiVavny5Lr30UknS0qVL1aNHD23dulUDBw7UunXrtGvXLm3YsEFJSUnq06ePZs+ercmTJ2vGjBmKiIhociz0vgEA8EBFRYXbVlNT87Pn1NbW6p///KduvPFGWSwW5efnq66uThkZGa4x3bt3V4cOHZSXlydJysvLU8+ePd3a+pmZmaqoqNDOnTs9iplkDwAwBacP/iNJqampstlsrm3u3Lk/+92rV69WWVmZrr/+eklScXGxIiIiFB8f7zYuKSlJxcXFrjH/m+gbjzce8wRtfACAKTgMQw7j9Nv4jecWFRXJarW69kdGRv7suUuWLNGQIUOUkpJy2t/vDZI9AAAesFqtbsn+5xw4cEAbNmzQyy+/7NqXnJys2tpalZWVuVX3JSUlSk5Odo354IMP3K5VUlLiOuYJ2vgAAFNoXKDnzXY6li5dqsTERA0dOtS1r1+/fgoPD9fGjRtd+woKClRYWCi73S5Jstvt2rFjh0pLS11j1q9fL6vVqvT0dI9ioLIHAJiCU4YcLbgaX2p4rPvSpUuVlZXldm+8zWbT2LFjNXHiRCUkJMhqtWr8+PGy2+0aOHCgJGnw4MFKT0/X6NGjNW/ePBUXF2vKlCnKzs5u0tTB/yLZAwDQTDZs2KDCwkLdeOONJx2bP3++QkJCNHLkSLeH6jQKDQ3VmjVrNG7cONntdsXExCgrK0uzZs3yOA6LYXixWsHPKioqZLPZ9O3nnXmCHoJWZkoff4cANJt6o06b9G+Vl5d7NA/uicZcsfezZMV5kSuOHXPqnO7FzRprc6GyBwCYgq9W4wciymEAAIIclT0AwBSc323enB+oSPYAAFNweLka35tz/Y1kDwAwBYchL99657tYWhpz9gAABDkqewCAKTBnDwBAkHPKIocsXp0fqGjjAwAQ5KjsAQCm4DQaNm/OD1QkewCAKTi8bON7c66/0cYHACDIUdkDAEzBzJU9yR4AYApOwyKn4cVqfC/O9Tfa+AAABDkqewCAKdDGBwAgyDkUIocXDW2HD2NpaSR7AIApGF7O2RvM2QMAgDMVlT0AwBSYswcAIMg5jBA5DC/m7AP4cbm08QEACHJU9gAAU3DKIqcXNa5TgVvak+wBAKZg5jl72vgAAAQ5KnsAgCl4v0CPNj4AAGe0hjl7L16EQxsfAACcqajsAQCm4PTy2fisxgcA4AzHnD0AAEHOqRDT3mfPnD0AAEGOyh4AYAoOwyKHF6+p9eZcfyPZAwBMweHlAj0HbXwAAHCmorIHAJiC0wiR04vV+E5W4wMAcGajjQ8AAIIWlT0AwBSc8m5FvdN3obQ4kj0AwBS8f6hO4DbDAzdyAADQJFT2AABT8P7Z+IFbH5PsAQCmYOb32ZPsAQCmYObKPnAjBwAATUJlDwAwBe8fqhO49THJHgBgCk7DIqc399kH8FvvAvfXFAAA0CRU9gAAU3B62cYP5IfqkOwBAKbg/VvvAjfZB27kAACgSajsAQCm4JBFDi8ejOPNuf5GsgcAmAJtfAAAELSo7AEApuCQd614h+9CaXEkewCAKZi5jU+yBwCYAi/CAQAAPnfo0CFdd911atOmjVq1aqWePXtq27ZtruOGYWjatGk6++yz1apVK2VkZOiLL75wu8bRo0c1atQoWa1WxcfHa+zYsaqsrPQoDpI9AMAUjO/eZ3+6m+HhfP+3336rCy64QOHh4Xr99de1a9cuPfjgg2rdurVrzLx587RgwQItXrxY77//vmJiYpSZmanq6mrXmFGjRmnnzp1av3691qxZo7ffflt//vOfPYqFNj4AwBR81cavqKhw2x8ZGanIyMiTxv/f//2fUlNTtXTpUte+Tp06uf7ZMAw9/PDDmjJlioYNGyZJevbZZ5WUlKTVq1frmmuu0e7du7V27Vp9+OGH6t+/vyTp0Ucf1ZVXXqm///3vSklJaVLsVPYAAHggNTVVNpvNtc2dO/eU4/7zn/+of//++sMf/qDExET17dtXTz75pOv4/v37VVxcrIyMDNc+m82mAQMGKC8vT5KUl5en+Ph4V6KXpIyMDIWEhOj9999vcsxU9gAAU/DVK26LiopktVpd+09V1UvSvn37tGjRIk2cOFF//etf9eGHH+q2225TRESEsrKyVFxcLElKSkpyOy8pKcl1rLi4WImJiW7Hw8LClJCQ4BrTFCR7AIApOLx8613juVar1S3Z/xin06n+/ftrzpw5kqS+ffvq008/1eLFi5WVlXXacZwO2vgAADSDs88+W+np6W77evToocLCQklScnKyJKmkpMRtTElJietYcnKySktL3Y7X19fr6NGjrjFNQbIHAJhCYxvfm80TF1xwgQoKCtz2ff7550pLS5PUsFgvOTlZGzdudB2vqKjQ+++/L7vdLkmy2+0qKytTfn6+a8ybb74pp9OpAQMGNDkW2vgAAFNwKkROL2pcT8+dMGGCfv3rX2vOnDm6+uqr9cEHH+iJJ57QE088IUmyWCy64447dO+99+rcc89Vp06dNHXqVKWkpGj48OGSGjoBV1xxhW6++WYtXrxYdXV1ysnJ0TXXXNPklfgSyR4AgGZx/vnna9WqVbrnnns0a9YsderUSQ8//LBGjRrlGnPXXXepqqpKf/7zn1VWVqbf/OY3Wrt2raKiolxjnnvuOeXk5Oiyyy5TSEiIRo4cqQULFngUi8UwDMNnP1kLq6iokM1m07efd5Y1jhkJBKfMlD7+DgFoNvVGnTbp3yovL2/SorfT0Zgrxr0zQpGx4ad9nZrKOi268OVmjbW5UNkDAEzBV7feBSKSPQDAFAwv33pn8CIcAABwpqKyBwCYgkMWOTx8mc0Pzw9UJHsAgCk4De/m3Z0Bu5ydNj4AAEGPyt7kxvwqXSUHI07af1XWV8qZe0i11RY9MTNFm/7TWnU1FvW75JjGzz2o1m3rXWMLtrfS03NS9MV/o2WxGOrW57jGTjmsc35RfdJ1AX84b0Cl/nDrVzq353G1Sa7XjBs7Km+tzW1MapdqjZ1yRL0GVio0TDrweaRm39xRXx1q+PdjyKhvNOh336pLzxOKiXNqRPfzVFUR6o8fB6fJ6eUCPW/O9bfAjRw+seD1Av1r+6eube6KPZKkC68qlyQtntFOW9fbNOXxL/X3l/foaEm4Zo3t6Dr/RFWI/jbqHLVNqdUjaz7Xg6v3qFWsU3/70zmqr/PHTwScLCraqX07o/TYX9uf8vjZaTV6aPUeFe2J1KTfn6NbLuuq5Q8nqbb6+5ZvVCuntm2K04pHE095DZz5nLJ4vQWqMyLZ5+bmqmPHjoqKitKAAQP0wQcf+Dsk04hv41BCYr1re3+DTWd3rFEve6WqKkL0xr8S9P9mHFKf31Tq3F4nNPGhQu3aFqvd+dGSpKI9kTr2bZjGTCpWapcadexWresmFuvbr8JP2TEA/GHbW1Y9M+9sbflBNd/o+ruL9cGbVi25N0V7P43WkQOR2rrOpvJvvn8Ay6qn2mrlY0n6LD+mpcIGfMbvyf7555/XxIkTNX36dH300Ufq3bu3MjMzT3rLD5pfXa1Fb77UWpnXfCOLRfriv9GqrwtR3wsrXWM6nFujxHa12v3dX3jtz6mRtXW93vhXG9XVWlRzwqK1/2qjDudWKzm11l8/CtBkFouhX11WoUP7InXf8r16/r879ciaL2S/otzfocHHHIbF6y1Q+T3ZP/TQQ7r55pt1ww03KD09XYsXL1Z0dLSefvppf4dmOlvW2lRZEarBVx+VJB0tDVN4hFOxNofbuPi2dTpa2rDcIzrWqQde2qONL7fWbzv30vBze2nbW3G697m9CmVFCAJA/Fn1io516o85pdr2llX3XNtZ7621atpTX6rnwMqfvwACRuOcvTdboPJr5LW1tcrPz1dGRoZrX0hIiDIyMpSXl3fS+JqaGlVUVLht8J03/pWg8wdVqE1y/c8P/k7NCYse+kuqfnF+lR5e87ke+vcX6ti9WlNHd1bNicD9LRjmYfnub8G8N6xa9WRb7dvZSisfS9L7G6waOuYb/wYH+Ihfk/3XX38th8OhpKQkt/1JSUkqLi4+afzcuXNls9lcW2pqakuFGvRKDobr43fidMWfvv/LLSGxXnW1Iaosd19xXPZVuBISG34heGtVa5UURegv8wvVrc8J9eh3XHfnHlBxYYTy3jj1/ChwJqk4Gqr6OunA51Fu+4u+iFRiO6aigolTXr7PngV6LeOee+5ReXm5aysqKvJ3SEFj3Yo2ij+rXgMyvu+WnNvruMLCnfr43VjXvqI9kSo9FKEe/aokSTUnQhQSIln+59+BkBBDFovkdLZY+MBpq68L0eefRKv9OTVu+9t1rlEpi0yDiuHlSnwjgJO9X2dVzzrrLIWGhqqkpMRtf0lJiZKTk08aHxkZqcjIyJYKzzScTmnd8wnK+MNRt3n2GKtTmdce1RMz2iku3qGYOIdy/9ZePfpVqUe/45Kkvhcd05P3puixv7bXsBu/ktNp0crHEhUaJvW+gPlOnBmioh1K6fR9lZ6cWqvOvzihY2Wh+upQhF5YmKi/Lj6gT7fG6JMtseo/6JgGXl6hSb8/x3VO67Z1ap1Yr5RODb8UdOp+QserQvXVoXAdK2OBSiDgrXd+EhERoX79+mnjxo0aPny4JMnpdGrjxo3KycnxZ2im8vHbcSo9FKHMa46edOyWGYcUYjE0++aOqquxqP8lx5Qz96DreIdzazRz2T4991Cy7riqqywhhrqcd0L3PbdXbZKaPvcPNKeuvU/ogZf2uj7fMvOwJGnd86314IQO2rLWpgV3t9M1OaUaN/uQDu5reKDOzg++72oNHfONRv/l+8LkwdUN1/v7HalavzKhhX4S4PRYDMPw69N+n3/+eWVlZenxxx/Xr371Kz388MNauXKlPvvss5Pm8n+ooqJCNptN337eWda4gJqRAJosM6WPv0MAmk29UadN+rfKy8tltVqb5Tsac8Xv1t+g8JjTn5qpq6rVqsuXNmuszcXvvac//vGP+uqrrzRt2jQVFxerT58+Wrt27c8megAAPEEb389ycnJo2wMA0EzOiGQPAEBz8/b59oF86x3JHgBgCmZu47OqDQCAIEdlDwAwBTNX9iR7AIApmDnZ08YHACDIUdkDAEzBzJU9yR4AYAqGvLt9zq+Pm/USyR4AYApmruyZswcAIMhR2QMATMHMlT3JHgBgCmZO9rTxAQAIclT2AABTMHNlT7IHAJiCYVhkeJGwvTnX32jjAwAQ5KjsAQCmwPvsAQAIcmaes6eNDwBAkKOyBwCYgpkX6JHsAQCmYOY2PskeAGAKZq7smbMHACDIUdkDAEzB8LKNH8iVPckeAGAKhiTD8O78QEUbHwCAIEdlDwAwBacssvAEPQAAgher8QEAQNCisgcAmILTsMjCQ3UAAAhehuHlavwAXo5PGx8AgCBHZQ8AMAUzL9Aj2QMATIFkDwBAkDPzAj3m7AEACHJU9gAAU2A1PgAAQa4h2Vu82Dz7vhkzZshisbht3bt3dx2vrq5Wdna22rRpo9jYWI0cOVIlJSVu1ygsLNTQoUMVHR2txMRETZo0SfX19R7/7FT2AAA0k1/84hfasGGD63NY2Pdpd8KECXr11Vf1wgsvyGazKScnRyNGjNB7770nSXI4HBo6dKiSk5O1ZcsWHTlyRGPGjFF4eLjmzJnjURwkewCAKfhjNX5YWJiSk5NP2l9eXq4lS5Zo+fLluvTSSyVJS5cuVY8ePbR161YNHDhQ69at065du7RhwwYlJSWpT58+mj17tiZPnqwZM2YoIiKiyXHQxgcAmILhg02SKioq3Laampof/c4vvvhCKSkp6ty5s0aNGqXCwkJJUn5+vurq6pSRkeEa2717d3Xo0EF5eXmSpLy8PPXs2VNJSUmuMZmZmaqoqNDOnTs9+tlJ9gAAeCA1NVU2m821zZ0795TjBgwYoGXLlmnt2rVatGiR9u/frwsvvFDHjh1TcXGxIiIiFB8f73ZOUlKSiouLJUnFxcVuib7xeOMxT9DGBwCYgq/a+EVFRbJara79kZGRpxw/ZMgQ1z/36tVLAwYMUFpamlauXKlWrVqddhyng8oeAGAOPurjW61Wt+3Hkv0PxcfHq2vXrtqzZ4+Sk5NVW1ursrIytzElJSWuOf7k5OSTVuc3fj7VOoCfQrIHAJiDV7fdWSQvn6BXWVmpvXv36uyzz1a/fv0UHh6ujRs3uo4XFBSosLBQdrtdkmS327Vjxw6Vlpa6xqxfv15Wq1Xp6ekefTdtfAAAmsGdd96pq666SmlpaTp8+LCmT5+u0NBQXXvttbLZbBo7dqwmTpyohIQEWa1WjR8/Xna7XQMHDpQkDR48WOnp6Ro9erTmzZun4uJiTZkyRdnZ2U3uJjQi2QMATKGln6B38OBBXXvttfrmm2/Utm1b/eY3v9HWrVvVtm1bSdL8+fMVEhKikSNHqqamRpmZmVq4cKHr/NDQUK1Zs0bjxo2T3W5XTEyMsrKyNGvWLI9jJ9kDAEyhpe+zX7FixU8ej4qKUm5urnJzc390TFpaml577TWPvvdUmLMHACDIUdkDAMzB20V2AfyKW5I9AMAUeOsdAAAIWlT2AABz+N8H3J/u+QGKZA8AMAV/vPXuTNGkZP+f//ynyRf87W9/e9rBAAAA32tSsh8+fHiTLmaxWORwOLyJBwCA5hPArXhvNCnZO53O5o4DAIBmZeY2vler8aurq30VBwAAzctHb70LRB4ne4fDodmzZ6tdu3aKjY3Vvn37JElTp07VkiVLfB4gAADwjsfJ/r777tOyZcs0b948RUREuPafd955euqpp3waHAAAvmPxwRaYPE72zz77rJ544gmNGjVKoaGhrv29e/fWZ5995tPgAADwGdr4TXfo0CF16dLlpP1Op1N1dXU+CQoAAPiOx8k+PT1d77zzzkn7X3zxRfXt29cnQQEA4HMmruw9foLetGnTlJWVpUOHDsnpdOrll19WQUGBnn32Wa1Zs6Y5YgQAwHsmfuudx5X9sGHD9Morr2jDhg2KiYnRtGnTtHv3br3yyiu6/PLLmyNGAADghdN6Nv6FF16o9evX+zoWAACajZlfcXvaL8LZtm2bdu/eLalhHr9fv34+CwoAAJ/jrXdNd/DgQV177bV67733FB8fL0kqKyvTr3/9a61YsULt27f3dYwAAMALHs/Z33TTTaqrq9Pu3bt19OhRHT16VLt375bT6dRNN93UHDECAOC9xgV63mwByuPKfvPmzdqyZYu6devm2tetWzc9+uijuvDCC30aHAAAvmIxGjZvzg9UHif71NTUUz48x+FwKCUlxSdBAQDgcyaes/e4jf/AAw9o/Pjx2rZtm2vftm3bdPvtt+vvf/+7T4MDAADea1Jl37p1a1ks389VVFVVacCAAQoLazi9vr5eYWFhuvHGGzV8+PBmCRQAAK+Y+KE6TUr2Dz/8cDOHAQBAMzNxG79JyT4rK6u54wAAAM3ktB+qI0nV1dWqra1122e1Wr0KCACAZmHiyt7jBXpVVVXKyclRYmKiYmJi1Lp1a7cNAIAzkonfeudxsr/rrrv05ptvatGiRYqMjNRTTz2lmTNnKiUlRc8++2xzxAgAALzgcRv/lVde0bPPPqtLLrlEN9xwgy688EJ16dJFaWlpeu655zRq1KjmiBMAAO+YeDW+x5X90aNH1blzZ0kN8/NHjx6VJP3mN7/R22+/7dvoAADwkcYn6HmzBSqPk33nzp21f/9+SVL37t21cuVKSQ0Vf+OLcQAAwJnD42R/ww036JNPPpEk3X333crNzVVUVJQmTJigSZMm+TxAAAB8wsQL9Dyes58wYYLrnzMyMvTZZ58pPz9fXbp0Ua9evXwaHAAA8J5X99lLUlpamtLS0nwRCwAAzcYiL99657NIWl6Tkv2CBQuafMHbbrvttIMBAAC+16RkP3/+/CZdzGKx+CXZ/65rT4VZwlv8e4GWsH8F02MIXs7j1dIN/26ZLzPxrXdNSvaNq+8BAAhYPC4XAAAEK68X6AEAEBBMXNmT7AEApuDtU/BM9QQ9AAAQWKjsAQDmYOI2/mlV9u+8846uu+462e12HTp0SJL0j3/8Q++++65PgwMAwGdM/Lhcj5P9Sy+9pMzMTLVq1Uoff/yxampqJEnl5eWaM2eOzwMEAADe8TjZ33vvvVq8eLGefPJJhYd//yCbCy64QB999JFPgwMAwFfM/Ipbj+fsCwoKdNFFF52032azqayszBcxAQDgeyZ+gp7HlX1ycrL27Nlz0v53331XnTt39klQAAD4HHP2TXfzzTfr9ttv1/vvvy+LxaLDhw/rueee05133qlx48Y1R4wAAMALHrfx7777bjmdTl122WU6fvy4LrroIkVGRurOO+/U+PHjmyNGAAC8ZuaH6nic7C0Wi/72t79p0qRJ2rNnjyorK5Wenq7Y2NjmiA8AAN8w8X32p/1QnYiICKWnp/syFgAA0Aw8TvaDBg2SxfLjKxLffPNNrwICAKBZeHv7nJkq+z59+rh9rqur0/bt2/Xpp58qKyvLV3EBAOBbtPGbbv78+afcP2PGDFVWVnodEAAA8C2fvfXuuuuu09NPP+2rywEA4FvcZ++9vLw8RUVF+epyAAD4lD8fl3v//ffLYrHojjvucO2rrq5Wdna22rRpo9jYWI0cOVIlJSVu5xUWFmro0KGKjo5WYmKiJk2apPr6eo+/3+M2/ogRI9w+G4ahI0eOaNu2bZo6darHAQAAEMw+/PBDPf744+rVq5fb/gkTJujVV1/VCy+8IJvNppycHI0YMULvvfeeJMnhcGjo0KFKTk7Wli1bdOTIEY0ZM0bh4eEev3jO48reZrO5bQkJCbrkkkv02muvafr06Z5eDgCAgFJRUeG2Nb799VQqKys1atQoPfnkk2rdurVrf3l5uZYsWaKHHnpIl156qfr166elS5dqy5Yt2rp1qyRp3bp12rVrl/75z3+qT58+GjJkiGbPnq3c3FzV1tZ6FLNHlb3D4dANN9ygnj17ugUNAMAZz0er8VNTU912T58+XTNmzDjlKdnZ2Ro6dKgyMjJ07733uvbn5+errq5OGRkZrn3du3dXhw4dlJeXp4EDByovL089e/ZUUlKSa0xmZqbGjRunnTt3qm/fvk0O3aNkHxoaqsGDB2v37t0kewBAQPHV43KLiopktVpd+yMjI085fsWKFfroo4/04YcfnnSsuLhYERERio+Pd9uflJSk4uJi15j/TfSNxxuPecLjOfvzzjtP+/btU6dOnTw9FQCAgGe1Wt2S/akUFRXp9ttv1/r168+Ixesez9nfe++9uvPOO7VmzRodOXLkpLkLAADOWC10211+fr5KS0v1y1/+UmFhYQoLC9PmzZu1YMEChYWFKSkpSbW1tSorK3M7r6SkRMnJyZIaXin/w9X5jZ8bxzRVk5P9rFmzVFVVpSuvvFKffPKJfvvb36p9+/Zq3bq1Wrdurfj4eFr7AIAzVwveZ3/ZZZdpx44d2r59u2vr37+/Ro0a5frn8PBwbdy40XVOQUGBCgsLZbfbJUl2u107duxQaWmpa8z69etltVo9fjdNk9v4M2fO1C233KK33nrLoy8AAMBs4uLidN5557nti4mJUZs2bVz7x44dq4kTJyohIUFWq1Xjx4+X3W7XwIEDJUmDBw9Wenq6Ro8erXnz5qm4uFhTpkxRdnb2j64T+DFNTvaG0fArzcUXX+zRFwAAcCY4095nP3/+fIWEhGjkyJGqqalRZmamFi5c6DoeGhqqNWvWaNy4cbLb7YqJiVFWVpZmzZrl8Xd5tEDvp952BwDAGc3PL8LZtGmT2+eoqCjl5uYqNzf3R89JS0vTa6+95t0Xy8Nk37Vr159N+EePHvUqIAAA4FseJfuZM2fKZrM1VywAADSbM62N35I8SvbXXHONEhMTmysWAACaj4nfZ9/kW++YrwcAIDB5vBofAICAZOLKvsnJ3ul0NmccAAA0K+bsAQAIdiau7D1+Nj4AAAgsVPYAAHMwcWVPsgcAmIKZ5+xp4wMAEOSo7AEA5kAbHwCA4EYbHwAABC0qewCAOdDGBwAgyJk42dPGBwAgyFHZAwBMwfLd5s35gYpkDwAwBxO38Un2AABT4NY7AAAQtKjsAQDmQBsfAAATCOCE7Q3a+AAABDkqewCAKZh5gR7JHgBgDiaes6eNDwBAkKOyBwCYAm18AACCHW18AAAQrKjsAQCmQBsfAIBgZ+I2PskeAGAOJk72zNkDABDkqOwBAKbAnD0AAMGONj4AAAhWVPYAAFOwGIYsxumX596c628kewCAOdDGBwAAwYrKHgBgCqzGBwAg2NHGBwAAwYrKHgBgCrTxAQAIdiZu45PsAQCmYObKnjl7AACCHJU9AMAcaOMDABD8ArkV7w3a+AAABDkqewCAORhGw+bN+QGKZA8AMAVW4wMAgKBFZQ8AMAdW4wMAENwszobNm/MDFW18AACCHMkeOm9ApWY+s1/LP9qpNw5/IvsV5SeNSe1SrRnL9uvlz3bo33t2aMFrn6ttu1rX8SGjvtG8F/fo5YIdeuPwJ4qxOlryRwCazPbvUnW65r9KeOawa19YcY0SH/xSHW7eqbQbPlXbhw8opKzO7bzEB/YrNXu30kbvUOotu9T2sUKFHq374eVxJjN8sHlg0aJF6tWrl6xWq6xWq+x2u15//XXX8erqamVnZ6tNmzaKjY3VyJEjVVJS4naNwsJCDR06VNHR0UpMTNSkSZNUX1/v8Y9Osoeiop3atzNKj/21/SmPn51Wo4dW71HRnkhN+v05uuWyrlr+cJJqqy3fX6OVU9s2xWnFo4ktFTbgsYi9xxW34RvVdIhy7bNUO5U8Z78k6cjUzjo88xxZ6g0lP/Cl5Pz+b/fq9FiV3tFBhx7qptIJaQorqVXi/AMt/SPAC42r8b3ZPNG+fXvdf//9ys/P17Zt23TppZdq2LBh2rlzpyRpwoQJeuWVV/TCCy9o8+bNOnz4sEaMGOE63+FwaOjQoaqtrdWWLVv0zDPPaNmyZZo2bZrHP7tf5+zffvttPfDAA8rPz9eRI0e0atUqDR8+3J8hmdK2t6za9pb1R49ff3exPnjTqiX3prj2HTkQ6TZm1VNtJUm97JXNEyTgJUu1Q4mPFurrP7dX/Mulrv1RBVUK+6pWh+4/V0Z0qCTpq1tTlTZ2p6J2Vqq6Z5wkqWJoW9c59W0jVD6srRIfPCDVG1KYRQgALXyf/VVXXeX2+b777tOiRYu0detWtW/fXkuWLNHy5ct16aWXSpKWLl2qHj16aOvWrRo4cKDWrVunXbt2acOGDUpKSlKfPn00e/ZsTZ48WTNmzFBERESTY/FrZV9VVaXevXsrNzfXn2HgJ1gshn51WYUO7YvUfcv36vn/7tQja744ZasfOJO1efqwjve1upK3S70hWSQj/PuEbYRbJIsU9VnVKa8VUlmvmHfLVNM1mkRvQhUVFW5bTU3Nz57jcDi0YsUKVVVVyW63Kz8/X3V1dcrIyHCN6d69uzp06KC8vDxJUl5ennr27KmkpCTXmMzMTFVUVLi6A03l18p+yJAhGjJkSJPH19TUuP2PWlFR0Rxh4X/En1Wv6Fin/phTqmX/l6wl96Wo/6AKTXvqS931+3O0Y2usv0MEflbMljJF7j+hw/d1OelYzbnRMiJDlLC8WN9ekywZUut/HZHFKYWWuc+Ntn7uiKzrvlZIjaHqc6NVclfHFvoJ4Au+eqhOamqq2/7p06drxowZpzxnx44dstvtqq6uVmxsrFatWqX09HRt375dERERio+PdxuflJSk4uJiSVJxcbFbom883njMEwF1693cuXM1c+ZMf4dhKpbvej95b1i16smGNua+na2U3v+4ho75hmSPM17o17Vq88xhHflrJxkRJzczndYwld6RpjZLDsm69mvJIlX9Ol41nVpJPyjay69qq2ODEhT2da1av1SitguLGhK+heo+IPjoPvuioiJZrd9PfUZGRv7ICVK3bt20fft2lZeX68UXX1RWVpY2b97sRRCnJ6CS/T333KOJEye6PldUVJz0GxZ8q+JoqOrrpAOfR7ntL/oiUr/41albnMCZJHL/CYWW16vdPV+49lmcDS166xtf68t/9tSJ3nE6uKC7QirqpVCLnDGhSv1/u1T/a5vbtZzWMDmtYapPiVRpu0h1yP5MkV8cV03XmJb+seBHjavrmyIiIkJdujR0lPr166cPP/xQjzzyiP74xz+qtrZWZWVlbtV9SUmJkpOTJUnJycn64IMP3K7XuFq/cUxTBVSyj4yM/MnfoOB79XUh+vyTaLU/x31Oql3nGpUebPriEMBfTpwXq4MPdHXb13ZRkepSIlU2LFEK+b4qd1ob/kqM+rRSoRX1Ot7vJ/5C/67Ks9QF8GPVTOZMeDa+0+lUTU2N+vXrp/DwcG3cuFEjR46UJBUUFKiwsFB2u12SZLfbdd9996m0tFSJiQ13Oq1fv15Wq1Xp6ekefW9AJXs0j6hoh1I6fX/PfHJqrTr/4oSOlYXqq0MRemFhov66+IA+3RqjT7bEqv+gYxp4eYUm/f4c1zmt29apdWK9Ujo1/FLQqfsJHa8K1VeHwnWsjD9m8B+jVajqUkPd9jkjQ+SIC1NdakPHKnbTUdW1i5QjLkyRXxxXm2cOq+LKs1SX0nA88ovjith7XDXdY+SMCVVYSa1aryxWXVKEqrtGt/jPhNPUwqvx77nnHg0ZMkQdOnTQsWPHtHz5cm3atElvvPGGbDabxo4dq4kTJyohIUFWq1Xjx4+X3W7XwIEDJUmDBw9Wenq6Ro8erXnz5qm4uFhTpkxRdna2x4UvfwtDXXuf0AMv7XV9vmVmw8NG1j3fWg9O6KAta21acHc7XZNTqnGzD+ngvkjNvrmjdn7w/Xz90DHfaPRfvn8YxIOrG6739ztStX5lQgv9JMDpCT9co9b/KlZopUP1bcNV9rtEVVx5luu4M9KimA/L1frFEllqnHLEh+lE7ziVjUiSwnlcCU6ttLRUY8aM0ZEjR2Sz2dSrVy+98cYbuvzyyyVJ8+fPV0hIiEaOHKmamhplZmZq4cKFrvNDQ0O1Zs0ajRs3Tna7XTExMcrKytKsWbM8jsViGP57QW9lZaX27NkjSerbt68eeughDRo0SAkJCerQocPPnl9RUSGbzaZLNExhlvDmDhfwi/0revk7BKDZOI9Xa/8Nc1ReXt7keXBPNeYK+5BZCguP+vkTfkR9XbXyXp/WrLE2F79W9tu2bdOgQYNcnxsX32VlZWnZsmV+igoAEJR4651/XHLJJfJjYwEAAFNgzh4AYApnwmp8fyHZAwDMwWm4vdzotM4PUCR7AIA5mHjOnntGAAAIclT2AABTsMjLOXufRdLySPYAAHNo4SfonUlo4wMAEOSo7AEApsCtdwAABDtW4wMAgGBFZQ8AMAWLYcjixSI7b871N5I9AMAcnN9t3pwfoGjjAwAQ5KjsAQCmQBsfAIBgZ+LV+CR7AIA58AQ9AAAQrKjsAQCmwBP0AAAIdrTxAQBAsKKyBwCYgsXZsHlzfqAi2QMAzIE2PgAACFZU9gAAc+ChOgAABDczPy6XNj4AAEGOyh4AYA4mXqBHsgcAmIMh795JH7i5nmQPADAH5uwBAEDQorIHAJiDIS/n7H0WSYsj2QMAzMHEC/Ro4wMAEOSo7AEA5uCUZPHy/ABFsgcAmAKr8QEAQNCisgcAmIOJF+iR7AEA5mDiZE8bHwCAIEdlDwAwBxNX9iR7AIA5cOsdAADBjVvvAABA0KKyBwCYA3P2AAAEOachWbxI2M7ATfa08QEACHJU9gAAc6CNDwBAsPMy2Stwkz1tfAAAghyVPQDAHGjjAwAQ5JyGvGrFsxofAACcqajsAQDmYDgbNm/OD1AkewCAOTBnDwBAkGPOHgAA+NLcuXN1/vnnKy4uTomJiRo+fLgKCgrcxlRXVys7O1tt2rRRbGysRo4cqZKSErcxhYWFGjp0qKKjo5WYmKhJkyapvr7eo1hI9gAAc2hs43uzeWDz5s3Kzs7W1q1btX79etXV1Wnw4MGqqqpyjZkwYYJeeeUVvfDCC9q8ebMOHz6sESNGuI47HA4NHTpUtbW12rJli5555hktW7ZM06ZN8ygW2vgAAHMw5OWcfcN/VVRUuO2OjIxUZGTkScPXrl3r9nnZsmVKTExUfn6+LrroIpWXl2vJkiVavny5Lr30UknS0qVL1aNHD23dulUDBw7UunXrtGvXLm3YsEFJSUnq06ePZs+ercmTJ2vGjBmKiIhoUuhU9gAAeCA1NVU2m821zZ07t0nnlZeXS5ISEhIkSfn5+aqrq1NGRoZrTPfu3dWhQwfl5eVJkvLy8tSzZ08lJSW5xmRmZqqiokI7d+5scsxU9gAAc/DRavyioiJZrVbX7lNV9T/kdDp1xx136IILLtB5550nSSouLlZERITi4+PdxiYlJam4uNg15n8TfePxxmNNRbIHAJiD0ynJi3vlnQ3nWq1Wt2TfFNnZ2fr000/17rvvnv73e4E2PgAAzSgnJ0dr1qzRW2+9pfbt27v2Jycnq7a2VmVlZW7jS0pKlJyc7Brzw9X5jZ8bxzQFyR4AYA4tvBrfMAzl5ORo1apVevPNN9WpUye34/369VN4eLg2btzo2ldQUKDCwkLZ7XZJkt1u144dO1RaWuoas379elmtVqWnpzc5Ftr4AABzaOEn6GVnZ2v58uX697//rbi4ONccu81mU6tWrWSz2TR27FhNnDhRCQkJslqtGj9+vOx2uwYOHChJGjx4sNLT0zV69GjNmzdPxcXFmjJlirKzs5u0VqARyR4AgGawaNEiSdIll1zitn/p0qW6/vrrJUnz589XSEiIRo4cqZqaGmVmZmrhwoWusaGhoVqzZo3GjRsnu92umJgYZWVladasWR7FQrIHAJhDCz8u12hCJyAqKkq5ubnKzc390TFpaWl67bXXPPruHyLZAwBMwTCcMrx4c5035/obyR4AYA6G4d3LbAL4rXesxgcAIMhR2QMAzMHwcs4+gCt7kj0AwBycTsnixbx7AM/Z08YHACDIUdkDAMyBNj4AAMHNcDpleNHGD+Rb72jjAwAQ5KjsAQDmQBsfAIAg5zQkizmTPW18AACCHJU9AMAcDEOSN/fZB25lT7IHAJiC4TRkeNHGb8pb7M5UJHsAgDkYTnlX2XPrHQAAOENR2QMATIE2PgAAwc7EbfyATvaNv2XVq86r5yQAZzLn8Wp/hwA0G+eJGkktUzV7myvqVee7YFqYxQjgvsTBgweVmprq7zAAAF4qKipS+/btm+Xa1dXV6tSpk4qLi72+VnJysvbv36+oqCgfRNZyAjrZO51OHT58WHFxcbJYLP4OxxQqKiqUmpqqoqIiWa1Wf4cD+BR/vlueYRg6duyYUlJSFBLSfGvGq6urVVtb6/V1IiIiAi7RSwHexg8JCWm23wTx06xWK38ZImjx57tl2Wy2Zv+OqKiogEzSvsKtdwAABDmSPQAAQY5kD49ERkZq+vTpioyM9HcogM/x5xvBKqAX6AEAgJ9HZQ8AQJAj2QMAEORI9gAABDmSPQAAQY5kjybLzc1Vx44dFRUVpQEDBuiDDz7wd0iAT7z99tu66qqrlJKSIovFotWrV/s7JMCnSPZokueff14TJ07U9OnT9dFHH6l3797KzMxUaWmpv0MDvFZVVaXevXsrNzfX36EAzYJb79AkAwYM0Pnnn6/HHntMUsN7CVJTUzV+/Hjdfffdfo4O8B2LxaJVq1Zp+PDh/g4F8Bkqe/ys2tpa5efnKyMjw7UvJCREGRkZysvL82NkAICmINnjZ3399ddyOBxKSkpy25+UlOSTV0YCAJoXyR4AgCBHssfPOuussxQaGqqSkhK3/SUlJUpOTvZTVACApiLZ42dFRESoX79+2rhxo2uf0+nUxo0bZbfb/RgZAKApwvwdAALDxIkTlZWVpf79++tXv/qVHn74YVVVVemGG27wd2iA1yorK7Vnzx7X5/3792v79u1KSEhQhw4d/BgZ4Bvceocme+yxx/TAAw+ouLhYffr00YIFCzRgwAB/hwV4bdOmTRo0aNBJ+7OysrRs2bKWDwjwMZI9AABBjjl7AACCHMkeAIAgR7IHACDIkewBAAhyJHsAAIIcyR4AgCBHsgcAIMiR7AEACHIke8BL119/vYYPH+76fMkll+iOO+5o8Tg2bdoki8WisrKyHx1jsVi0evXqJl9zxowZ6tOnj1dxffnll7JYLNq+fbtX1wFw+kj2CErXX3+9LBaLLBaLIiIi1KVLF82aNUv19fXN/t0vv/yyZs+e3aSxTUnQAOAtXoSDoHXFFVdo6dKlqqmp0Wuvvabs7GyFh4frnnvuOWlsbW2tIiIifPK9CQkJPrkOAPgKlT2CVmRkpJKTk5WWlqZx48YpIyND//nPfyR933q/7777lJKSom7dukmSioqKdPXVVys+Pl4JCQkaNmyYvvzyS9c1HQ6HJk6cqPj4eLVp00Z33XWXfvh6iR+28WtqajR58mSlpqYqMjJSXbp00ZIlS/Tll1+6Xr7SunVrWSwWXX/99ZIaXiE8d+5cderUSa1atVLv3r314osvun3Pa6+9pq5du6pVq1YaNGiQW5xNNXnyZHXt2lXR0dHq3Lmzpk6dqrq6upPGPf7440pNTVV0dLSuvvpqlZeXux1/6qmn1KNHD0VFRal79+5auHChx7EAaD4ke5hGq1atVFtb6/q8ceNGFRQUaP369VqzZo3q6uqUmZmpuLg4vfPOO3rvvfcUGxurK664wnXegw8+qGXLlunpp5/Wu+++q6NHj2rVqlU/+b1jxozRv/71Ly1YsEC7d+/W448/rtjYWKWmpuqll16SJBUUFOjIkSN65JFHJElz587Vs88+q8WLF2vnzp2aMGGCrrvuOm3evFlSwy8lI0aM0FVXXaXt27frpptu0t133+3x/yZxcXFatmyZdu3apUceeURPPvmk5s+f7zZmz549WrlypV555RWtXbtWH3/8sW699VbX8eeee07Tpk3Tfffdp927d2vOnDmaOnWqnnnmGY/jAdBMDCAIZWVlGcOGDTMMwzCcTqexfv16IzIy0rjzzjtdx5OSkoyamhrXOf/4xz+Mbt26GU6n07WvpqbGaNWqlfHGG28YhmEYZ599tjFv3jzX8bq6OqN9+/au7zIMw7j44ouN22+/3TAMwygoKDAkGevXrz9lnG+99ZYhyfj2229d+6qrq43o6Ghjy5YtbmPHjh1rXHvttYZhGMY999xjpKenux2fPHnySdf6IUnGqlWrfvT4Aw88YPTr18/1efr06UZoaKhx8OBB177XX3/dCAkJMY4cOWIYhmGcc845xvLly92uM3v2bMNutxuGYRj79+83JBkff/zxj34vgObFnD2C1po1axQbG6u6ujo5nU796U9/0owZM1zHe/bs6TZP/8knn2jPnj2Ki4tzu051dbX27t2r8vJyHTlyRAMGDHAdCwsLU//+/U9q5Tfavn27QkNDdfHFFzc57j179uj48eO6/PLL3fbX1taqb9++kqTdu3e7xSFJdru9yd/R6Pnnn9eCBQu0d+9eVVZWqr6+Xlar1W1Mhw4d1K5dO7fvcTqdKigoUFxcnPbu3auxY8fq5ptvdo2pr6+XzWbzOB4AzYNkj6A1aNAgLVq0SBEREUpJSVFYmPsf95iYGLfPlZWV6tevn5577rmTrtW2bdvTiqFVq1Yen1NZWSlJevXVV92SrNSwDsFX8vLyNGrUKM2cOVOZmZmy2WxasWKFHnzwQY9jffLJJ0/65SM0NNRnsQLwDskeQSsmJkZdunRp8vhf/vKXev7555WYmHhSddvo7LPP1vvvv6+LLrpIUkMFm5+fr1/+8penHN+zZ085nU5t3rxZGRkZJx1v7Cw4HA7XvvT0dEVGRqqwsPBHOwI9evRwLTZstHXr1p//If/Hli1blJaWpr/97W+ufQcOHDhpXGFhoQ4fPqyUlBTX94SEhKhbt25KSkpSSkqK9u3bp1GjRnn0/QBaDgv0gO+MGjVKZ511loYNG6Z33nlH+/fv16ZNm3Tbbbfp4MGDkqTbb79d999/v1avXq3PPvtMt95660/eI9+xY0dlZWXpxhtv1OrVq13XXLlypSQpLS1NFotFa9as0VdffaXKykrFxcXpzjvv1IQJE/TMM89o7969+uijj/Too4+6Fr3dcsst+uKLLzRp0iQVFBRo+fLlWrZsmUc/77nnnqvCwkKtWLFCe/fu1YIFC0652DAqKkpZWVn65JNP9M477+i2227T1VdfreTkZEnSzJkzNXfuXC1YsECff/65duzYoaVLl+qhhx7yKB4AzYdkD3wnOjpab7/9tjp06KARI0aoR48eGjt2rKqrq12V/l/+8heNHj1aWVlZstvtiouL0+9+97ufvO6iRYv0+9//Xrfeequ6d++um2++WVVVVZKkdu3aaebMmbr77ruVlJSknJwcSdLs2bM1depUzZ07Vz169NAVV1yhV199VZ06dZLUMI/+0ksvafXq1erdu7cWL16sOXPmePTz/va3v9WECROUk5OjPn36aMuWLZo6depJ47p06aIRI0boyiuv1ODBg9WrVy+3W+tuuukmPfXUU1q6dKl69uypiy++WMuWLXPFCsD/LMaPrSwCAABBgcoeAIAgR7IHACDIkewBAAhyJHsAAIIcyR4AgCBHsgcAIMiR7AEACHIkewAAghzJHgCAIEeyBwAgyJHsAQAIcv8ftuD10hKK16AAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_test_pred_classes = (\n",
    "    pd.DataFrame(model_nn.predict(tf.data.Dataset.from_tensor_slices((X_test)).batch(batch_size=batch_size), verbose=0)\n",
    "    .flatten())\n",
    "    .applymap(lambda x: 1 if x>0.5 else 0)\n",
    "    .rename(columns={0:'pred'})\n",
    ")\n",
    "confusion_matrix_model = ConfusionMatrixDisplay(confusion_matrix(np.array(dataframe_test.target), np.array(y_test_pred_classes.pred)))\n",
    "confusion_matrix_model.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ed684339-b5c5-4eba-8b5f-e8aed78f02ab",
   "metadata": {
    "tags": []
   },
   "source": [
    "### We have lost a significant amount of true negatives but gained an even higher amount of true positives."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "92938a83-8f2b-429e-aae8-6231ef705f80",
   "metadata": {},
   "source": [
    "<a id='decision_tree'/>\n",
    "\n",
    "## Decision tree classifier\n",
    "[Back to TOC](#TOC)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "111b82ca-b06c-40f9-adf8-cf44e260552c",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "id": "b182cf67-c937-4343-9b27-583e1bbbf813",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "stemmer = nltk.PorterStemmer()\n",
    "cv_analyzer = TfidfVectorizer(min_df=2).build_analyzer()\n",
    "\n",
    "def stemmed_words(doc):\n",
    "    return (stemmer.stem(w) for w in cv_analyzer(doc))\n",
    "\n",
    "vect = TfidfVectorizer(min_df=2, analyzer = stemmed_words)\n",
    "\n",
    "# transform the text in the training data to a text-term matrix\n",
    "X_train_vectorized = vect.fit_transform(X_train)\n",
    "X_test_vectorized = vect.transform(X_test)\n",
    "X_val_vectorized = vect.transform(X_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 216,
   "id": "058e978e-e8ae-4df7-ac1f-0a57d9c0f68b",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-49 {color: black;background-color: white;}#sk-container-id-49 pre{padding: 0;}#sk-container-id-49 div.sk-toggleable {background-color: white;}#sk-container-id-49 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-49 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-49 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-49 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-49 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-49 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-49 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-49 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-49 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-49 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-49 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-49 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-49 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-49 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-49 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-49 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-49 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-49 div.sk-item {position: relative;z-index: 1;}#sk-container-id-49 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-49 div.sk-item::before, #sk-container-id-49 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-49 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-49 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-49 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-49 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-49 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-49 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-49 div.sk-label-container {text-align: center;}#sk-container-id-49 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-49 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-49\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>DecisionTreeClassifier(max_depth=70, min_samples_leaf=7)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-74\" type=\"checkbox\" checked><label for=\"sk-estimator-id-74\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">DecisionTreeClassifier</label><div class=\"sk-toggleable__content\"><pre>DecisionTreeClassifier(max_depth=70, min_samples_leaf=7)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "DecisionTreeClassifier(max_depth=70, min_samples_leaf=7)"
      ]
     },
     "execution_count": 216,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tree_model = DecisionTreeClassifier(\n",
    "        criterion = \"gini\", \n",
    "        max_depth = 70, min_samples_leaf = 7) \n",
    "\n",
    "\n",
    "\n",
    "\n",
    "# train \n",
    "tree_model.fit(X_train_vectorized, y_train) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 217,
   "id": "fd11a1aa-5b39-47cc-9f1d-1a129c186389",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy = 0.68089\n",
      "AUC = 0.709\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7fda975cec50>"
      ]
     },
     "execution_count": 217,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "predictions = tree_model.predict(X_test_vectorized)\n",
    "predict_probab = tree_model.predict_proba(X_test_vectorized)[:,1]\n",
    "\n",
    "print(\"Accuracy = {:.5f}\".format(accuracy_score(y_test, predictions)))\n",
    "print(\"AUC = {:.3f}\".format(roc_auc_score(y_test, predict_probab)))\n",
    "\n",
    "\n",
    "confusion_matrix_model = ConfusionMatrixDisplay(confusion_matrix(y_test, predictions))\n",
    "confusion_matrix_model.plot()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c7e19cd6-2b8e-40d5-8aa9-0bb8f6b1a869",
   "metadata": {
    "tags": []
   },
   "source": [
    "### This is by far the worst model. The true negatives are quite high but the true positives are underestimated."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "557b94f4-a2bc-4fe7-b135-4573bcbfdf5d",
   "metadata": {},
   "source": [
    "<a id='random_forest'/>\n",
    "\n",
    "## Random Forest\n",
    "[Back to TOC](#TOC)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 218,
   "id": "930bc3e1-1da0-4f2b-807a-ca98f7e0059d",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-50 {color: black;background-color: white;}#sk-container-id-50 pre{padding: 0;}#sk-container-id-50 div.sk-toggleable {background-color: white;}#sk-container-id-50 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-50 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-50 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-50 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-50 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-50 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-50 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-50 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-50 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-50 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-50 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-50 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-50 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-50 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-50 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-50 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-50 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-50 div.sk-item {position: relative;z-index: 1;}#sk-container-id-50 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-50 div.sk-item::before, #sk-container-id-50 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-50 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-50 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-50 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-50 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-50 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-50 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-50 div.sk-label-container {text-align: center;}#sk-container-id-50 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-50 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-50\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>RandomForestClassifier(n_estimators=300, n_jobs=-1, random_state=777)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-75\" type=\"checkbox\" checked><label for=\"sk-estimator-id-75\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">RandomForestClassifier</label><div class=\"sk-toggleable__content\"><pre>RandomForestClassifier(n_estimators=300, n_jobs=-1, random_state=777)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "RandomForestClassifier(n_estimators=300, n_jobs=-1, random_state=777)"
      ]
     },
     "execution_count": 218,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create the model with 100 trees\n",
    "random_forest_model = RandomForestClassifier(n_estimators=300, \n",
    "                               random_state=777, \n",
    "                               max_features = 'sqrt',\n",
    "                               n_jobs=-1, verbose = 0)\n",
    "\n",
    "# train \n",
    "random_forest_model.fit(X_train_vectorized, y_train) \n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "id": "321d076a-ba23-4c6f-b216-e2cbf26311a6",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy = 0.77741\n",
      "AUC = 0.827\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7fda9749e750>"
      ]
     },
     "execution_count": 219,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "predictions = random_forest_model.predict(X_test_vectorized)\n",
    "predict_probab = random_forest_model.predict_proba(X_test_vectorized)[:,1]\n",
    "\n",
    "print(\"Accuracy = {:.5f}\".format(accuracy_score(y_test, predictions)))\n",
    "print(\"AUC = {:.3f}\".format(roc_auc_score(y_test, predict_probab)))\n",
    "\n",
    "\n",
    "confusion_matrix_model = ConfusionMatrixDisplay(confusion_matrix(y_test, predictions))\n",
    "confusion_matrix_model.plot()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 220,
   "id": "2092c41d-f48e-4bdd-94a7-21b92591d7cf",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train set Accuracy = 0.98768\n",
      "Train set AUC = 0.999\n"
     ]
    }
   ],
   "source": [
    "print(\"Train set Accuracy = {:.5f}\".format(accuracy_score(y_train, random_forest_model.predict(vect.transform(X_train)))))\n",
    "print(\"Train set AUC = {:.3f}\".format(roc_auc_score(y_train, random_forest_model.predict_proba(vect.transform(X_train))[:,1])))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3fecd808-8bbc-476f-be0a-256a1e3abcb8",
   "metadata": {},
   "source": [
    "### We see that the random forest classifier is overfitting the training set, however, it yields its best AUC score on the test set while it is of overfitting."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c8d3d28e-eb8e-4cf8-b7f9-1e7ebb1a50b2",
   "metadata": {},
   "source": [
    "<a id='knn'/>\n",
    "\n",
    "## KNN classifier\n",
    "[Back to TOC](#TOC)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 221,
   "id": "7e67042c-f49c-44c5-8624-30ce6641e1f8",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-51 {color: black;background-color: white;}#sk-container-id-51 pre{padding: 0;}#sk-container-id-51 div.sk-toggleable {background-color: white;}#sk-container-id-51 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-51 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-51 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-51 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-51 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-51 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-51 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-51 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-51 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-51 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-51 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-51 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-51 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-51 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-51 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-51 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-51 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-51 div.sk-item {position: relative;z-index: 1;}#sk-container-id-51 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-51 div.sk-item::before, #sk-container-id-51 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-51 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-51 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-51 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-51 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-51 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-51 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-51 div.sk-label-container {text-align: center;}#sk-container-id-51 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-51 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-51\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>KNeighborsClassifier(metric=&#x27;cosine&#x27;, n_neighbors=7)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-76\" type=\"checkbox\" checked><label for=\"sk-estimator-id-76\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">KNeighborsClassifier</label><div class=\"sk-toggleable__content\"><pre>KNeighborsClassifier(metric=&#x27;cosine&#x27;, n_neighbors=7)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "KNeighborsClassifier(metric='cosine', n_neighbors=7)"
      ]
     },
     "execution_count": 221,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "knn_model = KNeighborsClassifier(n_neighbors=7, metric='cosine')\n",
    "\n",
    "\n",
    "# train \n",
    "knn_model.fit(X_train_vectorized, y_train) #np.ravel(y_train)?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 222,
   "id": "2697495d-7e7b-464e-8edf-4cbdff9e5302",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy = 0.76822\n",
      "AUC = 0.803\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7fda9735b6d0>"
      ]
     },
     "execution_count": 222,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfsAAAGwCAYAAACuFMx9AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAAA5cklEQVR4nO3deXhU9dn/8c9kXydAJAmREEFkSWXRqDBV3BqJmkehUG0tQkTUXzEgQqFAFdmUWGyLooAbgrRS3AotiMqiAkrYglhkibJoIpCAYhICZps5vz9iRkdAMswkw8x5v67rXJdzzvecued5Uu657+93zrEYhmEIAAAErCBfBwAAABoXyR4AgABHsgcAIMCR7AEACHAkewAAAhzJHgCAAEeyBwAgwIX4OgBPOBwOHTx4ULGxsbJYLL4OBwDgJsMwdOzYMSUnJysoqPHqz8rKSlVXV3t8nbCwMEVERHghoqbl18n+4MGDSklJ8XUYAAAPFRUVqXXr1o1y7crKSrVNjVHxYbvH10pKStL+/fv9LuH7dbKPjY2VJH259QJZY5iRQGD6dYcuvg4BaDS1qtGHWu7897wxVFdXq/iwXV/mXyBr7NnnivJjDqWmf6Hq6mqSfVOqb91bY4I8+n8gcC4LsYT6OgSg8Xx/w/ammIqNibUoJvbs38ch/50u9utkDwBAQ9kNh+wePA3Gbji8F0wTI9kDAEzBIUMOnX229+RcX6P3DQBAgKOyBwCYgkMOedKI9+xs3yLZAwBMwW4Yshtn34r35Fxfo40PAECAo7IHAJiCmRfokewBAKbgkCG7SZM9bXwAAAIclT0AwBRo4wMAEOBYjQ8AAAIWlT0AwBQc32+enO+vSPYAAFOwe7ga35NzfY1kDwAwBbshD596571Ymhpz9gAABDgqewCAKTBnDwBAgHPIIrssHp3vr2jjAwAQ4KjsAQCm4DDqNk/O91ckewCAKdg9bON7cq6v0cYHACDAUdkDAEzBzJU9yR4AYAoOwyKH4cFqfA/O9TXa+AAABDgqewCAKdDGBwAgwNkVJLsHDW27F2NpaiR7AIApGB7O2RvM2QMAgB+74IILZLFYTtpycnIkSZWVlcrJyVF8fLxiYmLUv39/lZSUuFyjsLBQWVlZioqKUkJCgsaMGaPa2lq3Y6GyBwCYQlPP2W/evFl2+w/N/08//VQ33HCDbrvtNknSyJEj9dZbb+n1119XXFychg0bpn79+umjjz6qez+7XVlZWUpKStL69et16NAhDRo0SKGhoZo2bZpbsZDsAQCmYDeCZDc8mLN383a5LVu2dHn9+OOP68ILL9Q111yjsrIyzZ07VwsXLtT1118vSZo3b546d+6sDRs2qGfPnlqxYoV27typVatWKTExUd27d9fUqVM1duxYTZo0SWFhYQ2OhTY+AABuKC8vd9mqqqrOeE51dbX++c9/6u6775bFYlF+fr5qamqUkZHhHNOpUye1adNGeXl5kqS8vDx16dJFiYmJzjGZmZkqLy/Xjh073IqZZA8AMAWHLHIoyIOtro2fkpKiuLg455abm3vG916yZIlKS0t11113SZKKi4sVFhamZs2auYxLTExUcXGxc8yPE3398fpj7qCNDwAwBW/N2RcVFclqtTr3h4eHn/HcuXPn6qabblJycvJZv78nSPYAALjBarW6JPsz+fLLL7Vq1Sr9+9//du5LSkpSdXW1SktLXar7kpISJSUlOcds2rTJ5Vr1q/XrxzQUbXwAgCnUL9DzZDsb8+bNU0JCgrKyspz70tPTFRoaqtWrVzv3FRQUqLCwUDabTZJks9m0fft2HT582Dlm5cqVslqtSktLcysGKnsAgCnUzdl78CCcszjX4XBo3rx5ys7OVkjIDyk3Li5OQ4YM0ahRo9SiRQtZrVYNHz5cNptNPXv2lCT17t1baWlpGjhwoKZPn67i4mI9/PDDysnJadDUwY+R7AEAaCSrVq1SYWGh7r777pOOzZgxQ0FBQerfv7+qqqqUmZmp2bNnO48HBwdr2bJlGjp0qGw2m6Kjo5Wdna0pU6a4HYfFMAw3fzl47igvL1dcXJy+/aydrLHMSCAwZSZ393UIQKOpNWr0gf6jsrIyt+bB3VGfK17/pJOiYoPP+jonjtl1W7fdjRprY6GyBwCYguc31fHb2phkDwAwh/rfy5/9+f6b7Ol9AwAQ4KjsAQCmYDcssnvwmFpPzvU1kj0AwBTsCpLdg4a2nTY+AAA4V1HZAwBMwWEEyeHBanwHq/EBADi30cYHAAABi8oeAGAKDnm2ot7hvVCaHMkeAGAKnt9Ux3+b4f4bOQAAaBAqewCAKXh+b3z/rY9J9gAAU/DF8+zPFSR7AIApmLmy99/IAQBAg1DZAwBMwfOb6vhvfUyyBwCYgsOwyOHJ7+z9+Kl3/vs1BQAANAiVPQDAFBwetvH9+aY6JHsAgCl4/tQ7/032/hs5AABoECp7AIAp2GWR3YMb43hyrq+R7AEApkAbHwAABCwqewCAKdjlWSve7r1QmhzJHgBgCmZu45PsAQCmwINwAABAwKKyBwCYguHh8+wNfnoHAMC5jTY+AAAIWFT2AABTMPMjbkn2AABTsHv41DtPzvU1/40cAAA0CJU9AMAUaOMDABDgHAqSw4OGtifn+pr/Rg4AABqEyh4AYAp2wyK7B614T871NZI9AMAUmLMHACDAGR4+9c7gDnoAAOBcRWUPADAFuyyye/AwG0/O9TWSPQDAFByGZ/PuDsOLwTQx2vgAAAQ4KnuTG3RFmkq+Cjtp/y3ZRzQs94AkaeeWKM3/Syvt3hql4GCp3S++07SFexUeWfc1d+FTidq0yqp9OyIVEmbo37u3N+lnAM7k4h4Vuu3+I7qoywnFJ9Vq0t0XKO+duB+NMDRoTIlu/P03irHatXNLtGaOa62D+8OdI85vV6V7JxxU2uXHFRJqaP+uCC2Y3kqfrI9p+g+Es+LwcIGeJ+f6mv9GDq+Y+XaB/rXtU+eWu2iPJKnXLWWS6hL9QwMuVPrVxzRz+eeaufwz3Tr4a1l+9JdTW23R1beUKiv7a198BOCMIqIc2rcjQs/8ufUpj9+ec0R97j6ip8e11oj/u0iVJ4I0beE+hYY7nGOmvLxPQcGGxt52oYbd2EH7dkZqyoL9at6ypqk+BjzkkMXjzV0HDhzQnXfeqfj4eEVGRqpLly7asmWL87hhGHrkkUfUqlUrRUZGKiMjQ59//rnLNY4ePaoBAwbIarWqWbNmGjJkiCoqKtyK45xI9rNmzdIFF1ygiIgI9ejRQ5s2bfJ1SKbRLN6uFgm1zm3jqji1uqBKXW11f0jPTTpffYcc0W+HH9YFHSuV0r5K19xaqrDwHyavBo0pVr/7jqhtp0pffQzgZ21536qXp7fSepdqvp6hvvcc0b+eSlTeu3HavytS0x9oo/jEGv3yxrovvdYWtWp9YbVeeyZB+3dF6uD+cL30WCtFRDl0AX/3OI1vv/1WV155pUJDQ/X2229r586d+tvf/qbmzZs7x0yfPl0zZ87Us88+q40bNyo6OlqZmZmqrPzh72rAgAHasWOHVq5cqWXLlmnt2rW677773IrF58n+1Vdf1ahRozRx4kRt3bpV3bp1U2Zmpg4fPuzr0Eynptqi995srszffSOLRSr9OkS7t0arWXytHrzlIv226y80ul97fbox2tehAl6T1KZa8Ym12rou1rnvxLFg7f44Sp3TT0iSyo8Gq2hPuDJu+1bhkXYFBRvKGviNvj0Sos//F+mr0OGm+jvoebJJUnl5uctWVVV1yvf7y1/+opSUFM2bN09XXHGF2rZtq969e+vCCy+UVFfVP/nkk3r44YfVp08fde3aVQsWLNDBgwe1ZMkSSdKuXbv0zjvv6MUXX1SPHj101VVX6emnn9aiRYt08ODBBn92nyf7v//977r33ns1ePBgpaWl6dlnn1VUVJReeuklX4dmOuvfiVNFebB6335UknToy7q5/H/8PUk3DfhGj72yT+27nNC4316oA/tOnucH/FGLhFpJUukR1yVMpUdC1CKhvkVv0bjfttOFF3+nJZ9/qmX7/6d+9x3RQwPaqqKMpU/+on7O3pNNklJSUhQXF+fccnNzT/l+//3vf3XZZZfptttuU0JCgi655BK98MILzuP79+9XcXGxMjIynPvi4uLUo0cP5eXlSZLy8vLUrFkzXXbZZc4xGRkZCgoK0saNGxv82X36V1pdXa38/HyNHz/euS8oKEgZGRnOD/pjVVVVLt+gysvLmyROs3j3Xy10+XXlik+q+8fP8f105c13fqPM39V9AWjf5Ttt+zBW7y6K191/PuSrUIEmZmjYtAMq/TpEf/x1e1VXWnTjHUc1ef4XeuDmi3T0cKivA0QTKioqktVqdb4ODw8/5bh9+/Zpzpw5GjVqlP785z9r8+bNeuCBBxQWFqbs7GwVFxdLkhITE13OS0xMdB4rLi5WQkKCy/GQkBC1aNHCOaYhfFrZf/3117Lb7T/7QX8sNzfX5dtUSkpKU4Ua8Eq+CtXH62J14++/ce6LT6xL+qkdXOckU9pX6vAB/nFDYDh6uK7maday1mV/s5a1ziTe/aoKXZFRrtyhqdq5OVp7tkfpmT+3VnWlRRnfd8Jw7nPI4rw//llt3y/Qs1qtLtvpkr3D4dCll16qadOm6ZJLLtF9992ne++9V88++2xTfmxJ50Ab3x3jx49XWVmZcysqKvJ1SAFjxaJ4NTuvVj0yfuiWJKZUKz6pWl/tdf1DPrAvXAmtWYGMwFBcGKZvSkJ0yVXHnPuiYuzqdMkJ7cqPkiSFR9a1uRwO13MdhkVB/ntTNdMxPFyJb7i5Gr9Vq1ZKS0tz2de5c2cVFhZKkpKSkiRJJSUlLmNKSkqcx5KSkk5aw1ZbW6ujR486xzSET9v45513noKDg3/2g/5YeHj4ab9B4ew5HNKKV1so47ajCv7RX4TFIv1m6BH9469Japf2ndr94juter2FivZG6OEXvnCOO/xVqI6VhujwgVA57NLeT+sWLCW3rVJk9E/+dQR8ICLKruS21c7XSSnVaveL73SsNFhHDoRpyYstdceIwzqwP1zFhWHK/lOxvikJda7e35UfrYqyYI15qkivzEhUVWWQbhrwjZJSqrVptfV0b4tzTFM/9e7KK69UQUGBy77PPvtMqampkqS2bdsqKSlJq1evVvfu3SXVTU9v3LhRQ4cOlSTZbDaVlpYqPz9f6enpkqT33ntPDodDPXr0aHAsPk32YWFhSk9P1+rVq9W3b19JdW2P1atXa9iwYb4MzVQ+XhurwwfCnPPyP9bv3iOqqbTo2Ynn61hpsNqlVSr3X3uVfMEP/3Au+GsrrXythfP1/b07SpKmv7FH3X7p3m9BgcbQodt3euLNvc7Xf5hct4p5xavN9beRbfTarJaKiHJoxPSvFGO1a8fmaD00oJ1qquqan+VHQ/TQ79vprnGH9JfX9io41NCXBRGaNPgC7dvJanyc2siRI/XLX/5S06ZN0+23365Nmzbp+eef1/PPPy9JslgsevDBB/Xoo4/qoosuUtu2bTVhwgQlJyc7c2Lnzp114403Otv/NTU1GjZsmH73u98pOTm5wbFYDMPw6d1+X331VWVnZ+u5557TFVdcoSeffFKvvfaadu/efdJc/k+Vl5crLi5O337WTtZYv5qRABosM7m7r0MAGk2tUaMP9B+VlZW5LHrzpvpc8euVgxUaffa/JKo5Xq3FN8xzK9Zly5Zp/Pjx+vzzz9W2bVuNGjVK9957r/O4YRiaOHGinn/+eZWWluqqq67S7Nmz1aFDB+eYo0ePatiwYVq6dKmCgoLUv39/zZw5UzExDb97o8+TvSQ988wzeuKJJ1RcXKzu3btr5syZDWpPkOxhBiR7BLKmTPZ9VtztcbL/T++XGjXWxnJO/EB02LBhtO0BAGgk50SyBwCgsZ3t/e1/fL6/ItkDAEyhqVfjn0uY6AYAIMBR2QMATMHMlT3JHgBgCmZO9rTxAQAIcFT2AABTMHNlT7IHAJiCIc9+PufzO9B5gGQPADAFM1f2zNkDABDgqOwBAKZg5sqeZA8AMAUzJ3va+AAABDgqewCAKZi5sifZAwBMwTAsMjxI2J6c62u08QEACHBU9gAAU+B59gAABDgzz9nTxgcAIMBR2QMATMHMC/RI9gAAUzBzG59kDwAwBTNX9szZAwAQ4KjsAQCmYHjYxvfnyp5kDwAwBUOSYXh2vr+ijQ8AQICjsgcAmIJDFlm4gx4AAIGL1fgAACBgUdkDAEzBYVhk4aY6AAAELsPwcDW+Hy/Hp40PAECAo7IHAJiCmRfokewBAKZAsgcAIMCZeYEec/YAAAQ4KnsAgCmYeTU+yR4AYAp1yd6TOXsvBtPEaOMDABDgqOwBAKbAanwAAAKcIc+eSe/HXXza+AAABDoqewCAKdDGBwAg0Jm4j08bHwBgDt9X9me7yc3KftKkSbJYLC5bp06dnMcrKyuVk5Oj+Ph4xcTEqH///iopKXG5RmFhobKyshQVFaWEhASNGTNGtbW1bn90KnsAABrJL37xC61atcr5OiTkh7Q7cuRIvfXWW3r99dcVFxenYcOGqV+/fvroo48kSXa7XVlZWUpKStL69et16NAhDRo0SKGhoZo2bZpbcZDsAQCm4Is76IWEhCgpKemk/WVlZZo7d64WLlyo66+/XpI0b948de7cWRs2bFDPnj21YsUK7dy5U6tWrVJiYqK6d++uqVOnauzYsZo0aZLCwsIaHAdtfACAKXjSwv/x4r7y8nKXraqq6rTv+fnnnys5OVnt2rXTgAEDVFhYKEnKz89XTU2NMjIynGM7deqkNm3aKC8vT5KUl5enLl26KDEx0TkmMzNT5eXl2rFjh1ufnWQPAIAbUlJSFBcX59xyc3NPOa5Hjx6aP3++3nnnHc2ZM0f79+9Xr169dOzYMRUXFyssLEzNmjVzOScxMVHFxcWSpOLiYpdEX3+8/pg7aOMDAMzhLBbZnXS+pKKiIlmtVufu8PDwUw6/6aabnP/dtWtX9ejRQ6mpqXrttdcUGRl59nGcBSp7AIAp1M/Ze7JJktVqddlOl+x/qlmzZurQoYP27NmjpKQkVVdXq7S01GVMSUmJc44/KSnppNX59a9PtQ7g55DsAQBoAhUVFdq7d69atWql9PR0hYaGavXq1c7jBQUFKiwslM1mkyTZbDZt375dhw8fdo5ZuXKlrFar0tLS3Hpv2vgAAHNo4pvqjB49WrfccotSU1N18OBBTZw4UcHBwbrjjjsUFxenIUOGaNSoUWrRooWsVquGDx8um82mnj17SpJ69+6ttLQ0DRw4UNOnT1dxcbEefvhh5eTkNLibUI9kDwAwhaa+Xe5XX32lO+64Q998841atmypq666Shs2bFDLli0lSTNmzFBQUJD69++vqqoqZWZmavbs2c7zg4ODtWzZMg0dOlQ2m03R0dHKzs7WlClT3I69Qcn+v//9b4MveOutt7odBAAAgWbRokU/ezwiIkKzZs3SrFmzTjsmNTVVy5cv9ziWBiX7vn37NuhiFotFdrvdk3gAAGg8fnx/e080KNk7HI7GjgMAgEZl5qfeebQav7Ky0ltxAADQuAwvbH7K7WRvt9s1depUnX/++YqJidG+ffskSRMmTNDcuXO9HiAAAPCM28n+scce0/z58zV9+nSXm/BffPHFevHFF70aHAAA3mPxwuaf3E72CxYs0PPPP68BAwYoODjYub9bt27avXu3V4MDAMBraOM33IEDB9S+ffuT9jscDtXU1HglKAAA4D1uJ/u0tDStW7fupP1vvPGGLrnkEq8EBQCA15m4snf7DnqPPPKIsrOzdeDAATkcDv373/9WQUGBFixYoGXLljVGjAAAeM5LT73zR25X9n369NHSpUu1atUqRUdH65FHHtGuXbu0dOlS3XDDDY0RIwAA8MBZ3Ru/V69eWrlypbdjAQCg0fz4MbVne76/OusH4WzZskW7du2SVDePn56e7rWgAADwuiZ+6t25xO1kX/8Un48++kjNmjWTJJWWluqXv/ylFi1apNatW3s7RgAA4AG35+zvuece1dTUaNeuXTp69KiOHj2qXbt2yeFw6J577mmMGAEA8Fz9Aj1PNj/ldmW/Zs0arV+/Xh07dnTu69ixo55++mn16tXLq8EBAOAtFqNu8+R8f+V2sk9JSTnlzXPsdruSk5O9EhQAAF5n4jl7t9v4TzzxhIYPH64tW7Y4923ZskUjRozQX//6V68GBwAAPNegyr558+ayWH6Yqzh+/Lh69OihkJC602traxUSEqK7775bffv2bZRAAQDwiIlvqtOgZP/kk082chgAADQyE7fxG5Tss7OzGzsOAADQSM76pjqSVFlZqerqapd9VqvVo4AAAGgUJq7s3V6gd/z4cQ0bNkwJCQmKjo5W8+bNXTYAAM5JJn7qndvJ/k9/+pPee+89zZkzR+Hh4XrxxRc1efJkJScna8GCBY0RIwAA8IDbbfylS5dqwYIFuvbaazV48GD16tVL7du3V2pqql555RUNGDCgMeIEAMAzJl6N73Zlf/ToUbVr105S3fz80aNHJUlXXXWV1q5d693oAADwkvo76Hmy+Su3k327du20f/9+SVKnTp302muvSaqr+OsfjAMAAM4dbif7wYMH65NPPpEkjRs3TrNmzVJERIRGjhypMWPGeD1AAAC8wsQL9Nyesx85cqTzvzMyMrR7927l5+erffv26tq1q1eDAwAAnvPod/aSlJqaqtTUVG/EAgBAo7HIw6feeS2SptegZD9z5swGX/CBBx4462AAAID3NSjZz5gxo0EXs1gsPkn2//f/BiokJKLJ3xdoCl9NC/N1CECjcVRWSpP/0zRvZuKf3jUo2devvgcAwG9xu1wAABCoPF6gBwCAXzBxZU+yBwCYgqd3wTPVHfQAAIB/obIHAJiDidv4Z1XZr1u3TnfeeadsNpsOHDggSfrHP/6hDz/80KvBAQDgNSa+Xa7byf7NN99UZmamIiMj9fHHH6uqqkqSVFZWpmnTpnk9QAAA4Bm3k/2jjz6qZ599Vi+88IJCQ0Od+6+88kpt3brVq8EBAOAtZn7Erdtz9gUFBbr66qtP2h8XF6fS0lJvxAQAgPeZ+A56blf2SUlJ2rNnz0n7P/zwQ7Vr184rQQEA4HXM2TfcvffeqxEjRmjjxo2yWCw6ePCgXnnlFY0ePVpDhw5tjBgBAIAH3G7jjxs3Tg6HQ7/61a904sQJXX311QoPD9fo0aM1fPjwxogRAACPmfmmOm4ne4vFooceekhjxozRnj17VFFRobS0NMXExDRGfAAAeIeJf2d/1jfVCQsLU1pamjdjAQAAjcDtOfvrrrtO119//Wk3AADOSZ7+7M6Dyv7xxx+XxWLRgw8+6NxXWVmpnJwcxcfHKyYmRv3791dJSYnLeYWFhcrKylJUVJQSEhI0ZswY1dbWuv3+blf23bt3d3ldU1Ojbdu26dNPP1V2drbbAQAA0CR81MbfvHmznnvuOXXt2tVl/8iRI/XWW2/p9ddfV1xcnIYNG6Z+/frpo48+kiTZ7XZlZWUpKSlJ69ev16FDhzRo0CCFhoa6fRM7t5P9jBkzTrl/0qRJqqiocPdyAAD4lfLycpfX4eHhCg8PP+XYiooKDRgwQC+88IIeffRR5/6ysjLNnTtXCxcudHbF582bp86dO2vDhg3q2bOnVqxYoZ07d2rVqlVKTExU9+7dNXXqVI0dO1aTJk1SWFhYg2P22lPv7rzzTr300kveuhwAAN7lpd/Zp6SkKC4uzrnl5uae9i1zcnKUlZWljIwMl/35+fmqqalx2d+pUye1adNGeXl5kqS8vDx16dJFiYmJzjGZmZkqLy/Xjh073ProXnvqXV5eniIiIrx1OQAAvMpbP70rKiqS1Wp17j9dVb9o0SJt3bpVmzdvPulYcXGxwsLC1KxZM5f9iYmJKi4udo75caKvP15/zB1uJ/t+/fq5vDYMQ4cOHdKWLVs0YcIEdy8HAIBfsVqtLsn+VIqKijRixAitXLnynCiE3U72cXFxLq+DgoLUsWNHTZkyRb179/ZaYAAA+Kv8/HwdPnxYl156qXOf3W7X2rVr9cwzz+jdd99VdXW1SktLXar7kpISJSUlSaq7Pf2mTZtcrlu/Wr9+TEO5leztdrsGDx6sLl26qHnz5m69EQAAPtWEq/F/9atfafv27S77Bg8erE6dOmns2LFKSUlRaGioVq9erf79+0uqe9BcYWGhbDabJMlms+mxxx7T4cOHlZCQIElauXKlrFar2/e5cSvZBwcHq3fv3tq1axfJHgDgV5rydrmxsbG6+OKLXfZFR0crPj7euX/IkCEaNWqUWrRoIavVquHDh8tms6lnz56SpN69eystLU0DBw7U9OnTVVxcrIcfflg5OTmnXSdwOm638S+++GLt27dPbdu2dfdUAADwvRkzZigoKEj9+/dXVVWVMjMzNXv2bOfx4OBgLVu2TEOHDpXNZlN0dLSys7M1ZcoUt9/L7WT/6KOPavTo0Zo6darS09MVHR3tcvxMixYAAPAZH97f/oMPPnB5HRERoVmzZmnWrFmnPSc1NVXLly/3+L0bnOynTJmiP/7xj7r55pslSbfeeqssFovzuGEYslgsstvtHgcFAIDX8SCcM5s8ebL+8Ic/6P3332/MeAAAgJc1ONkbRt1XmmuuuabRggEAoLHwPPsG+nHbHgAAv0Ibv2E6dOhwxoR/9OhRjwICAADe5Vaynzx58kl30AMAwB/Qxm+g3/3ud867+AAA4FdM3MZv8CNuma8HAMA/ub0aHwAAv2Tiyr7Byd7hcDRmHAAANCrm7AEACHQmruwbPGcPAAD8E5U9AMAcTFzZk+wBAKZg5jl72vgAAAQ4KnsAgDnQxgcAILDRxgcAAAGLyh4AYA608QEACHAmTva08QEACHBU9gAAU7B8v3lyvr8i2QMAzMHEbXySPQDAFPjpHQAACFhU9gAAc6CNDwCACfhxwvYEbXwAAAIclT0AwBTMvECPZA8AMAcTz9nTxgcAIMBR2QMATIE2PgAAgY42PgAACFRU9gAAU6CNDwBAoDNxG59kDwAwBxMne+bsAQAIcFT2AABTYM4eAIBARxsfAAAEKip7AIApWAxDFuPsy3NPzvU1kj0AwBxo4wMAgEBFZQ8AMAUzr8ansgcAmIPhhc0Nc+bMUdeuXWW1WmW1WmWz2fT22287j1dWVionJ0fx8fGKiYlR//79VVJS4nKNwsJCZWVlKSoqSgkJCRozZoxqa2vd/ugkewAAGkHr1q31+OOPKz8/X1u2bNH111+vPn36aMeOHZKkkSNHaunSpXr99de1Zs0aHTx4UP369XOeb7fblZWVperqaq1fv14vv/yy5s+fr0ceecTtWGjjAwBMoanb+LfccovL68cee0xz5szRhg0b1Lp1a82dO1cLFy7U9ddfL0maN2+eOnfurA0bNqhnz55asWKFdu7cqVWrVikxMVHdu3fX1KlTNXbsWE2aNElhYWENjoXKHgBgDl5q45eXl7tsVVVVZ3xru92uRYsW6fjx47LZbMrPz1dNTY0yMjKcYzp16qQ2bdooLy9PkpSXl6cuXbooMTHROSYzM1Pl5eXO7kBDkewBAKZQX9l7sklSSkqK4uLinFtubu5p33P79u2KiYlReHi4/vCHP2jx4sVKS0tTcXGxwsLC1KxZM5fxiYmJKi4uliQVFxe7JPr64/XH3EEbHwAANxQVFclqtTpfh4eHn3Zsx44dtW3bNpWVlemNN95Qdna21qxZ0xRhuiDZAwDMwUs31alfXd8QYWFhat++vSQpPT1dmzdv1lNPPaXf/va3qq6uVmlpqUt1X1JSoqSkJElSUlKSNm3a5HK9+tX69WMaijY+AMA0PG3he8rhcKiqqkrp6ekKDQ3V6tWrnccKCgpUWFgom80mSbLZbNq+fbsOHz7sHLNy5UpZrValpaW59b5U9gAANILx48frpptuUps2bXTs2DEtXLhQH3zwgd59913FxcVpyJAhGjVqlFq0aCGr1arhw4fLZrOpZ8+ekqTevXsrLS1NAwcO1PTp01VcXKyHH35YOTk5Pzt1cCokewCAORhG3ebJ+W44fPiwBg0apEOHDikuLk5du3bVu+++qxtuuEGSNGPGDAUFBal///6qqqpSZmamZs+e7Tw/ODhYy5Yt09ChQ2Wz2RQdHa3s7GxNmTLF7dBJ9gAAU2jq39nPnTv3Z49HRERo1qxZmjVr1mnHpKamavny5e698SkwZw8AQICjsgcAmIOJH3FLsgcAmILFUbd5cr6/oo0PAECAo7I3uTv+7xP1uuxLtWlVqqqaEO34PEEvvHq5iorjTjHaUO4fV6hHtwOa8OSv9NHWVOeR9xa8dNLoqbOu1fsb2zVi9ID77uvysUanb9T8nV00bdOVkqSw4FqNvyxPN7fdo7Bguz48kKJJG3rpm8oo53m2Vl9pxCWb1aH5UX1XG6LFezpqxtYrZDeomfwGbXyYVbdOxfrPqs4q2H+egoIcuue2fE3/0zsaPK6fKqtDXcb+JnOHJMtpr/WX53tp0/bzna8rTjT8iUxAU+gSf1i/7bBTu4/Gu+z/8+XrdW3rQo34oLeO1YTpkR4f6pnr3tUdb/9aktSp+dd6IWO55vzvUv1p3fVKjDquKba1CrYY+ssWmy8+Cs5CU6/GP5f49Cvp2rVrdcsttyg5OVkWi0VLlizxZTimNO6vmXr3w4v0xYHm2lcUr7+80EuJ5x1Xh7bfuIy7sM03uu2mTzX9xatOe62KE2H6tizKudXU8F0S546okBr99erVmrD+GpVV//BFNCa0Sr+5aLdyN9u0ofh87fimpcZ/dK3SE0vUrWXdrUlvbrtXBd/Ga9Ynl6nwWJw2lyRr+paeGtDpU0WHVPvqI8Fd9b+z92TzUz5N9sePH1e3bt1+9jeGaFrRkTWSpPKKH+7OFB5Wq4eGrtFTC2z6tizqdKdqxKA8LZ71imZP/K9uvPoz+XXPCwFnYs91+uCrNlp/qLXL/ovjv1ZYsMNl/76y5jpQEaNLWtY9WSwsyK4qe7DLeVX2EEWE2PWL8440fvCAh3xaet1000266aabGjy+qqrK5bnB5eXljRGWaVkshnLu3KjtnyXoiwPNnfvv//1G7fg8Qet/NEf/Uy+9eak+3tlKVdUhuuziA3pwUJ4iw2u0eOUvmiJ04Gdltd2jtPiv1X9Zv5OOnRd5QtX2IB2rdr396DffReq8yO8kSesOpig7bbuy2n6ut7+4UC0jTyinW74kqWXkicb/APAKM7fx/arPmpubq8mTJ/s6jIA1YlCe2p7/rR54NMu575eXFOqStEO6b0Kfnz33n//p7vzvPV/GKyK8Vr+9+VOSPXwuKapCD13xkQav+D9V28/un7yPDqZo+paemmJbpyd6vadqe7Bm/y9dlycdkmGcfh0LzjEs0PMP48eP16hRo5yvy8vLlZKS4sOIAscDA/PUs3uRHnzsZn39bbRz/yVpB5WcUK6lz/7TZfykB97T9oJEjcq9+ZTX27W3pQb13abQELtqaoNPOQZoChefd0TnRX6nxbe84dwXEmTo8sRDurPTpxqyMkthwQ7FhlW5VPfxkd/p6+8ina/n7eymeTu7KiHyhMqqw9U65phGp29U0bGGPeoU8CW/Svbh4eFuP+kHZ2LogYEbdFX6lxqZe5OKv451ObpwWVe99UFHl30v5S7W7FeuUN7HbU571fZtvlF5RRiJHj6Xd/B8ZS253WXf41e9r31lzfT89kt06Hi0qu1BsrU6oBVf1v1UtK21VOfHVOjjIz99ZrhFh7+r+zKc1XaPDlbEaMfR85riY8ALaOPDtEZk5+lXPffp4Sd/pROVoWoeVzf/ePxEmKprQpwr63/q8Dcxzi8Gtu6Fah73nXbuSVB1TbAuu/iAfn/r//Ta8oub9LMAp3K8Nkyfl7Zw2XeiNkTfVkU497/xeSeNv3y9yqrCVVETpgk9PtTWw4n65Eii85whv9imdQdS5JBFvdvs131dPtaDa26Qg9/Z+48mfurduYRkb3J9frVbkvTkQ2+77P/L87307ocXNegatfYg9cnYpft/v1EWi3SgxKo5C684qSMAnKumbf6lDMOip69bobAguz48WHdTnR+7unWhhnbbqrAgu3Z/G6/737tRaw+cvrsFnEt8muwrKiq0Z88e5+v9+/dr27ZtatGihdq04X9ETeH6QXd7fM7m7a21eXvr04wGzj0D33FdcFptD9Hkjb00eWOv05whZb97a2OHhUZGG99HtmzZouuuu875un7xXXZ2tubPn++jqAAAAYnV+L5x7bXXyvDjORAAAPwBc/YAAFOgjQ8AQKBzGHWbJ+f7KZI9AMAcTDxnzw9EAQAIcFT2AABTsMjDOXuvRdL0SPYAAHMw8R30aOMDABDgqOwBAKbAT+8AAAh0rMYHAACBisoeAGAKFsOQxYNFdp6c62skewCAOTi+3zw530/RxgcAIMBR2QMATIE2PgAAgc7Eq/FJ9gAAc+AOegAAIFBR2QMATIE76AEAEOho4wMAgEBFZQ8AMAWLo27z5Hx/RbIHAJgDbXwAABCoqOwBAObATXUAAAhsZr5dLm18AAACHJU9AMAcTLxAj2QPADAHQ549k95/cz3JHgBgDszZAwAAr8rNzdXll1+u2NhYJSQkqG/fviooKHAZU1lZqZycHMXHxysmJkb9+/dXSUmJy5jCwkJlZWUpKipKCQkJGjNmjGpra92KhWQPADAHQz/M25/V5t7brVmzRjk5OdqwYYNWrlypmpoa9e7dW8ePH3eOGTlypJYuXarXX39da9as0cGDB9WvXz/ncbvdrqysLFVXV2v9+vV6+eWXNX/+fD3yyCNuxUIbHwBgDk28QO+dd95xeT1//nwlJCQoPz9fV199tcrKyjR37lwtXLhQ119/vSRp3rx56ty5szZs2KCePXtqxYoV2rlzp1atWqXExER1795dU6dO1dixYzVp0iSFhYU1KBYqewAA3FBeXu6yVVVVNei8srIySVKLFi0kSfn5+aqpqVFGRoZzTKdOndSmTRvl5eVJkvLy8tSlSxclJiY6x2RmZqq8vFw7duxocMwkewCAOTi8sElKSUlRXFycc8vNzT3zWzscevDBB3XllVfq4osvliQVFxcrLCxMzZo1cxmbmJio4uJi55gfJ/r64/XHGoo2PgDAFLy1Gr+oqEhWq9W5Pzw8/Izn5uTk6NNPP9WHH3541u/vCSp7AADcYLVaXbYzJfthw4Zp2bJlev/999W6dWvn/qSkJFVXV6u0tNRlfElJiZKSkpxjfro6v/51/ZiGINkDAMzBo5X47i/uMwxDw4YN0+LFi/Xee++pbdu2LsfT09MVGhqq1atXO/cVFBSosLBQNptNkmSz2bR9+3YdPnzYOWblypWyWq1KS0trcCy08QEA5tDEq/FzcnK0cOFC/ec//1FsbKxzjj0uLk6RkZGKi4vTkCFDNGrUKLVo0UJWq1XDhw+XzWZTz549JUm9e/dWWlqaBg4cqOnTp6u4uFgPP/ywcnJyGjR9UI9kDwBAI5gzZ44k6dprr3XZP2/ePN11112SpBkzZigoKEj9+/dXVVWVMjMzNXv2bOfY4OBgLVu2TEOHDpXNZlN0dLSys7M1ZcoUt2Ih2QMAzKGJK3ujAeMjIiI0a9YszZo167RjUlNTtXz5crfe+6dI9gAAc3BIsnh4vp8i2QMATIEH4QAAgIBFZQ8AMIcmnrM/l5DsAQDm4DAkiwcJ2+G/yZ42PgAAAY7KHgBgDrTxAQAIdB4me/lvsqeNDwBAgKOyBwCYA218AAACnMOQR614VuMDAIBzFZU9AMAcDEfd5sn5fopkDwAwB+bsAQAIcMzZAwCAQEVlDwAwB9r4AAAEOEMeJnuvRdLkaOMDABDgqOwBAOZAGx8AgADncEjy4LfyDv/9nT1tfAAAAhyVPQDAHGjjAwAQ4Eyc7GnjAwAQ4KjsAQDmYOLb5ZLsAQCmYBgOGR48uc6Tc32NZA8AMAfD8Kw6Z84eAACcq6jsAQDmYHg4Z+/HlT3JHgBgDg6HZPFg3t2P5+xp4wMAEOCo7AEA5kAbHwCAwGY4HDI8aOP780/vaOMDABDgqOwBAOZAGx8AgADnMCSLOZM9bXwAAAIclT0AwBwMQ5Inv7P338qeZA8AMAXDYcjwoI1vkOwBADjHGQ55Vtnz0zsAAHCOorIHAJgCbXwAAAKdidv4fp3s679l1dZW+TgSoPE4Kv33HxjgTBxVlZKapmquVY1H99SpVY33gmliFsOP+xJfffWVUlJSfB0GAMBDRUVFat26daNcu7KyUm3btlVxcbHH10pKStL+/fsVERHhhciajl8ne4fDoYMHDyo2NlYWi8XX4ZhCeXm5UlJSVFRUJKvV6utwAK/i77vpGYahY8eOKTk5WUFBjbdmvLKyUtXV1R5fJywszO8SveTnbfygoKBG+yaIn2e1WvnHEAGLv++mFRcX1+jvERER4ZdJ2lv46R0AAAGOZA8AQIAj2cMt4eHhmjhxosLDw30dCuB1/H0jUPn1Aj0AAHBmVPYAAAQ4kj0AAAGOZA8AQIAj2QMAEOBI9miwWbNm6YILLlBERIR69OihTZs2+TokwCvWrl2rW265RcnJybJYLFqyZImvQwK8imSPBnn11Vc1atQoTZw4UVu3blW3bt2UmZmpw4cP+zo0wGPHjx9Xt27dNGvWLF+HAjQKfnqHBunRo4cuv/xyPfPMM5LqnkuQkpKi4cOHa9y4cT6ODvAei8WixYsXq2/fvr4OBfAaKnucUXV1tfLz85WRkeHcFxQUpIyMDOXl5fkwMgBAQ5DscUZff/217Ha7EhMTXfYnJiZ65ZGRAIDGRbIHACDAkexxRuedd56Cg4NVUlLisr+kpERJSUk+igoA0FAke5xRWFiY0tPTtXr1auc+h8Oh1atXy2az+TAyAEBDhPg6APiHUaNGKTs7W5dddpmuuOIKPfnkkzp+/LgGDx7s69AAj1VUVGjPnj3O1/v379e2bdvUokULtWnTxoeRAd7BT+/QYM8884yeeOIJFRcXq3v37po5c6Z69Ojh67AAj33wwQe67rrrTtqfnZ2t+fPnN31AgJeR7AEACHDM2QMAEOBI9gAABDiSPQAAAY5kDwBAgCPZAwAQ4Ej2AAAEOJI9AAABjmQPAECAI9kDHrrrrrvUt29f5+trr71WDz74YJPH8cEHH8hisai0tPS0YywWi5YsWdLga06aNEndu3f3KK4vvvhCFotF27Zt8+g6AM4eyR4B6a677pLFYpHFYlFYWJjat2+vKVOmqLa2ttHf+9///remTp3aoLENSdAA4CkehIOAdeONN2revHmqqqrS8uXLlZOTo9DQUI0fP/6ksdXV1QoLC/PK+7Zo0cIr1wEAb6GyR8AKDw9XUlKSUlNTNXToUGVkZOi///2vpB9a74899piSk5PVsWNHSVJRUZFuv/12NWvWTC1atFCfPn30xRdfOK9pt9s1atQoNWvWTPHx8frTn/6knz5e4qdt/KqqKo0dO1YpKSkKDw9X+/btNXfuXH3xxRfOh680b95cFotFd911l6S6Rwjn5uaqbdu2ioyMVLdu3fTGG2+4vM/y5cvVoUMHRUZG6rrrrnOJs6HGjh2rDh06KCoqSu3atdOECRNUU1Nz0rjnnntOKSkpioqK0u23366ysjKX4y+++KI6d+6siIgIderUSbNnz3Y7FgCNh2QP04iMjFR1dbXz9erVq1VQUKCVK1dq2bJlqqmpUWZmpmJjY7Vu3Tp99NFHiomJ0Y033ug8729/+5vmz5+vl156SR9++KGOHj2qxYsX/+z7Dho0SP/61780c+ZM7dq1S88995xiYmKUkpKiN998U5JUUFCgQ4cO6amnnpIk5ebmasGCBXr22We1Y8cOjRw5UnfeeafWrFkjqe5LSb9+/XTLLbdo27ZtuueeezRu3Di3/28SGxur+fPna+fOnXrqqaf0wgsvaMaMGS5j9uzZo9dee01Lly7VO++8o48//lj333+/8/grr7yiRx55RI899ph27dqladOmacKECXr55ZfdjgdAIzGAAJSdnW306dPHMAzDcDgcxsqVK43w8HBj9OjRzuOJiYlGVVWV85x//OMfRseOHQ2Hw+HcV1VVZURGRhrvvvuuYRiG0apVK2P69OnO4zU1NUbr1q2d72UYhnHNNdcYI0aMMAzDMAoKCgxJxsqVK08Z5/vvv29IMr799lvnvsrKSiMqKspYv369y9ghQ4YYd9xxh2EYhjF+/HgjLS3N5fjYsWNPutZPSTIWL1582uNPPPGEkZ6e7nw9ceJEIzg42Pjqq6+c+95++20jKCjIOHTokGEYhnHhhRcaCxcudLnO1KlTDZvNZhiGYezfv9+QZHz88cenfV8AjYs5ewSsZcuWKSYmRjU1NXI4HPr973+vSZMmOY936dLFZZ7+k08+0Z49exQbG+tyncrKSu3du1dlZWU6dOiQevTo4TwWEhKiyy677KRWfr1t27YpODhY11xzTYPj3rNnj06cOKEbbrjBZX91dbUuueQSSdKuXbtc4pAkm83W4Peo9+qrr2rmzJnau3evKioqVFtbK6vV6jKmTZs2Ov/8813ex+FwqKCgQLGxsdq7d6+GDBmie++91zmmtrZWcXFxbscDoHGQ7BGwrrvuOs2ZM0dhYWFKTk5WSIjrn3t0dLTL64qKCqWnp+uVV1456VotW7Y8qxgiIyPdPqeiokKS9NZbb7kkWaluHYK35OXlacCAAZo8ebIyMzMVFxenRYsW6W9/+5vbsb7wwgsnffkIDg72WqwAPEOyR8CKjo5W+/btGzz+0ksv1auvvqqEhISTqtt6rVq10saNG3X11VdLqqtg8/Pzdemll55yfJcuXeRwOLRmzRplZGScdLy+s2C325370tLSFB4ersLCwtN2BDp37uxcbFhvw4YNZ/6QP7J+/XqlpqbqoYcecu778ssvTxpXWFiogwcPKjk52fk+QUFB6tixoxITE5WcnKx9+/ZpwIABbr0/gKbDAj3gewMGDNB5552nPn36aN26ddq/f78++OADPfDAA/rqq68kSSNGjNDjjz+uJUuWaPfu3br//vt/9jfyF1xwgbKzs3X33XdryZIlzmu+9tprkqTU1FRZLBYtW7ZMR44cUUVFhWJjYzV69GiNHDlSL7/8svbu3autW7fq6aefdi56+8Mf/qDPP/9cY8aMUUFBgRYuXKj58+e79XkvuugiFRYWatGiRdq7d69mzpx5ysWGERERys7O1ieffKJ169bpgQce0O23366kpCRJ0uTJk5Wbm6uZM2fqs88+0/bt2zVv3jz9/e9/dyseAI2HZA98LyoqSmvXrlWbNm3Ur18/de7cWUOGDFFlZaWz0v/jH/+ogQMHKjs7WzabTbGxsfr1r3/9s9edM2eOfvOb3+j+++9Xp06ddO+99+r48eOSpPPPP1+TJ0/WuHHjlJiYqGHDhkmSpk6dqgkTJig3N1edO3fWjTfeqLfeektt27aVVDeP/uabb2rJkiXq1q2bnn32WU2bNs2tz3vrrbdq5MiRGjZsmLp3767169drwoQJJ41r3769+vXrp5tvvlm9e/dW165dXX5ad8899+jFF1/UvHnz1KVLF11zzTWaP3++M1YAvmcxTreyCAAABAQqewAAAhzJHgCAAEeyBwAgwJHsAQAIcCR7AAACHMkeAIAAR7IHACDAkewBAAhwJHsAAAIcyR4AgABHsgcAIMD9f+WFAp9goa2IAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "predictions = knn_model.predict(X_test_vectorized)\n",
    "predict_probab = knn_model.predict_proba(X_test_vectorized)[:,1]\n",
    "\n",
    "print(\"Accuracy = {:.5f}\".format(accuracy_score(y_test, predictions)))\n",
    "print(\"AUC = {:.3f}\".format(roc_auc_score(y_test, predict_probab)))\n",
    "\n",
    "\n",
    "confusion_matrix_model = ConfusionMatrixDisplay(confusion_matrix(y_test, predictions))\n",
    "confusion_matrix_model.plot()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "62ba6ad6-2654-4e0f-b99a-2118c2262878",
   "metadata": {},
   "source": [
    "<a id='adaboo'/>\n",
    "\n",
    "## AdaBoost classifier\n",
    "[Back to TOC](#TOC)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "id": "d23b1c3d-ad59-451e-89cf-587b41d2721d",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-53 {color: black;background-color: white;}#sk-container-id-53 pre{padding: 0;}#sk-container-id-53 div.sk-toggleable {background-color: white;}#sk-container-id-53 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-53 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-53 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-53 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-53 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-53 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-53 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-53 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-53 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-53 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-53 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-53 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-53 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-53 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-53 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-53 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-53 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-53 div.sk-item {position: relative;z-index: 1;}#sk-container-id-53 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-53 div.sk-item::before, #sk-container-id-53 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-53 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-53 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-53 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-53 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-53 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-53 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-53 div.sk-label-container {text-align: center;}#sk-container-id-53 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-53 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-53\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>AdaBoostClassifier(random_state=777)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-78\" type=\"checkbox\" checked><label for=\"sk-estimator-id-78\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">AdaBoostClassifier</label><div class=\"sk-toggleable__content\"><pre>AdaBoostClassifier(random_state=777)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "AdaBoostClassifier(random_state=777)"
      ]
     },
     "execution_count": 224,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "adaboo_model = AdaBoostClassifier(random_state=777)\n",
    "adaboo_model.fit(X_train_vectorized, y_train)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 225,
   "id": "268d4692-3691-405d-ad23-a50bc7563740",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy = 0.75181\n",
      "AUC = 0.799\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7fda97368310>"
      ]
     },
     "execution_count": 225,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "predictions = adaboo_model.predict(X_test_vectorized)\n",
    "predict_probab = adaboo_model.predict_proba(X_test_vectorized)[:,1]\n",
    "\n",
    "print(\"Accuracy = {:.5f}\".format(accuracy_score(y_test, predictions)))\n",
    "print(\"AUC = {:.3f}\".format(roc_auc_score(y_test, predict_probab)))\n",
    "\n",
    "\n",
    "confusion_matrix_model = ConfusionMatrixDisplay(confusion_matrix(y_test, predictions))\n",
    "confusion_matrix_model.plot()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7fb55024-7638-4675-a068-7ff99c1ed965",
   "metadata": {},
   "source": [
    "<a id='stacking'/>\n",
    "\n",
    "## Lets try stacking! Yey!\n",
    "[Back to TOC](#TOC)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 243,
   "id": "8f66b401-b0cd-4092-8a11-ce95cb165cc6",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-61 {color: black;background-color: white;}#sk-container-id-61 pre{padding: 0;}#sk-container-id-61 div.sk-toggleable {background-color: white;}#sk-container-id-61 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-61 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-61 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-61 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-61 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-61 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-61 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-61 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-61 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-61 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-61 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-61 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-61 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-61 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-61 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-61 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-61 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-61 div.sk-item {position: relative;z-index: 1;}#sk-container-id-61 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-61 div.sk-item::before, #sk-container-id-61 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-61 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-61 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-61 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-61 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-61 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-61 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-61 div.sk-label-container {text-align: center;}#sk-container-id-61 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-61 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-61\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>StackingClassifier(estimators=[(&#x27;log_reg_model&#x27;,\n",
       "                                LogisticRegression(max_iter=777)),\n",
       "                               (&#x27;random_forest_model&#x27;,\n",
       "                                RandomForestClassifier(n_estimators=300,\n",
       "                                                       n_jobs=-1,\n",
       "                                                       random_state=777)),\n",
       "                               (&#x27;adaboo_model&#x27;,\n",
       "                                AdaBoostClassifier(random_state=777))],\n",
       "                   final_estimator=LogisticRegression())</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-121\" type=\"checkbox\" ><label for=\"sk-estimator-id-121\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">StackingClassifier</label><div class=\"sk-toggleable__content\"><pre>StackingClassifier(estimators=[(&#x27;log_reg_model&#x27;,\n",
       "                                LogisticRegression(max_iter=777)),\n",
       "                               (&#x27;random_forest_model&#x27;,\n",
       "                                RandomForestClassifier(n_estimators=300,\n",
       "                                                       n_jobs=-1,\n",
       "                                                       random_state=777)),\n",
       "                               (&#x27;adaboo_model&#x27;,\n",
       "                                AdaBoostClassifier(random_state=777))],\n",
       "                   final_estimator=LogisticRegression())</pre></div></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><label>log_reg_model</label></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-122\" type=\"checkbox\" ><label for=\"sk-estimator-id-122\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">LogisticRegression</label><div class=\"sk-toggleable__content\"><pre>LogisticRegression(max_iter=777)</pre></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><label>random_forest_model</label></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-123\" type=\"checkbox\" ><label for=\"sk-estimator-id-123\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">RandomForestClassifier</label><div class=\"sk-toggleable__content\"><pre>RandomForestClassifier(n_estimators=300, n_jobs=-1, random_state=777)</pre></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><label>adaboo_model</label></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-124\" type=\"checkbox\" ><label for=\"sk-estimator-id-124\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">AdaBoostClassifier</label><div class=\"sk-toggleable__content\"><pre>AdaBoostClassifier(random_state=777)</pre></div></div></div></div></div></div></div></div><div class=\"sk-item\"><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><label>final_estimator</label></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-125\" type=\"checkbox\" ><label for=\"sk-estimator-id-125\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">LogisticRegression</label><div class=\"sk-toggleable__content\"><pre>LogisticRegression()</pre></div></div></div></div></div></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "StackingClassifier(estimators=[('log_reg_model',\n",
       "                                LogisticRegression(max_iter=777)),\n",
       "                               ('random_forest_model',\n",
       "                                RandomForestClassifier(n_estimators=300,\n",
       "                                                       n_jobs=-1,\n",
       "                                                       random_state=777)),\n",
       "                               ('adaboo_model',\n",
       "                                AdaBoostClassifier(random_state=777))],\n",
       "                   final_estimator=LogisticRegression())"
      ]
     },
     "execution_count": 243,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "estimators = [\n",
    "    ('log_reg_model', log_reg_model),\n",
    "    #('tree_model', tree_model),\n",
    "    ('random_forest_model', random_forest_model),\n",
    "    #('knn_model', knn_model),\n",
    "    ('adaboo_model',adaboo_model)\n",
    "]\n",
    "\n",
    "stacking_model = StackingClassifier(estimators = estimators, final_estimator = LogisticRegression())\n",
    "stacking_model.fit(X_train_vectorized, y_train)#.score(X_test, y_test)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 247,
   "id": "67c6694f-ef66-4a3a-b543-4d835b905b2e",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy = 0.78792\n",
      "AUC = 0.849\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7fda9ce066d0>"
      ]
     },
     "execution_count": 247,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "predictions = stacking_model.predict(vect.transform(X_test))\n",
    "predict_probab = stacking_model.predict_proba(vect.transform(X_test))[:,1]\n",
    "\n",
    "print(\"Accuracy = {:.5f}\".format(accuracy_score(y_test, predictions)))\n",
    "print(\"AUC = {:.3f}\".format(roc_auc_score(y_test, predict_probab)))\n",
    "\n",
    "\n",
    "confusion_matrix_model = ConfusionMatrixDisplay(confusion_matrix(y_test, predictions))\n",
    "confusion_matrix_model.plot()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f8b6ac1d-fe8d-498a-a901-e5fd22fde98a",
   "metadata": {},
   "source": [
    "<a id='voting'/>\n",
    "\n",
    "## And a voting method\n",
    "[Back to TOC](#TOC)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 248,
   "id": "7b495843-165f-4425-98ac-489fb83fbb27",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-63 {color: black;background-color: white;}#sk-container-id-63 pre{padding: 0;}#sk-container-id-63 div.sk-toggleable {background-color: white;}#sk-container-id-63 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-63 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-63 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-63 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-63 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-63 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-63 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-63 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-63 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-63 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-63 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-63 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-63 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-63 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-63 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-63 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-63 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-63 div.sk-item {position: relative;z-index: 1;}#sk-container-id-63 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-63 div.sk-item::before, #sk-container-id-63 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-63 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-63 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-63 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-63 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-63 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-63 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-63 div.sk-label-container {text-align: center;}#sk-container-id-63 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-63 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-63\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>VotingClassifier(estimators=[(&#x27;log_reg_model&#x27;,\n",
       "                              LogisticRegression(max_iter=777)),\n",
       "                             (&#x27;random_forest_model&#x27;,\n",
       "                              RandomForestClassifier(n_estimators=300,\n",
       "                                                     n_jobs=-1,\n",
       "                                                     random_state=777)),\n",
       "                             (&#x27;adaboo_model&#x27;,\n",
       "                              AdaBoostClassifier(random_state=777))])</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-130\" type=\"checkbox\" ><label for=\"sk-estimator-id-130\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">VotingClassifier</label><div class=\"sk-toggleable__content\"><pre>VotingClassifier(estimators=[(&#x27;log_reg_model&#x27;,\n",
       "                              LogisticRegression(max_iter=777)),\n",
       "                             (&#x27;random_forest_model&#x27;,\n",
       "                              RandomForestClassifier(n_estimators=300,\n",
       "                                                     n_jobs=-1,\n",
       "                                                     random_state=777)),\n",
       "                             (&#x27;adaboo_model&#x27;,\n",
       "                              AdaBoostClassifier(random_state=777))])</pre></div></div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><label>log_reg_model</label></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-131\" type=\"checkbox\" ><label for=\"sk-estimator-id-131\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">LogisticRegression</label><div class=\"sk-toggleable__content\"><pre>LogisticRegression(max_iter=777)</pre></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><label>random_forest_model</label></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-132\" type=\"checkbox\" ><label for=\"sk-estimator-id-132\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">RandomForestClassifier</label><div class=\"sk-toggleable__content\"><pre>RandomForestClassifier(n_estimators=300, n_jobs=-1, random_state=777)</pre></div></div></div></div></div></div><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label sk-toggleable\"><label>adaboo_model</label></div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-133\" type=\"checkbox\" ><label for=\"sk-estimator-id-133\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">AdaBoostClassifier</label><div class=\"sk-toggleable__content\"><pre>AdaBoostClassifier(random_state=777)</pre></div></div></div></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "VotingClassifier(estimators=[('log_reg_model',\n",
       "                              LogisticRegression(max_iter=777)),\n",
       "                             ('random_forest_model',\n",
       "                              RandomForestClassifier(n_estimators=300,\n",
       "                                                     n_jobs=-1,\n",
       "                                                     random_state=777)),\n",
       "                             ('adaboo_model',\n",
       "                              AdaBoostClassifier(random_state=777))])"
      ]
     },
     "execution_count": 248,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "voting_model = VotingClassifier(estimators = estimators, voting = 'hard')\n",
    "voting_model.fit(X_train_vectorized, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 249,
   "id": "ee2995f0-c777-4000-a336-e0b21104e367",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy = 0.78726\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7fda9d3cc150>"
      ]
     },
     "execution_count": 249,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "predictions = voting_model.predict(vect.transform(X_test))\n",
    "#predict_probab = voting_model.predict_proba(vect.transform(X_test))[:,1]\n",
    "\n",
    "print(\"Accuracy = {:.5f}\".format(accuracy_score(y_test, predictions)))\n",
    "#print(\"AUC = {:.3f}\".format(roc_auc_score(y_test, predict_probab)))\n",
    "\n",
    "\n",
    "confusion_matrix_model = ConfusionMatrixDisplay(confusion_matrix(y_test, predictions))\n",
    "confusion_matrix_model.plot()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f49c632b-ad61-4edf-a994-47fc1a5221eb",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "ab167a53-29b0-4588-ab14-2baa8b052a20",
   "metadata": {},
   "source": [
    "<a id='conclusion'/>\n",
    "\n",
    "## Conclusion:\n",
    "[Back to TOC](#TOC)\n",
    "\n",
    "#### The highest AUC was achieved (for now) after stacking a Logistic Regression model, Random Forest, and Adaboost. It seems to yield the best balance between true negatives and true positives."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "86aa1c55-adbd-494d-a170-3dc711f378a3",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}