Classical_Models/train_XGB.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "fossil-tender",
   "metadata": {},
   "outputs": [],
   "source": [
    "author_list = ['Charles Dickens','Jane Austen','Sir Arthur Conan Doyle','George Eliot','Jules Verne']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "veterinary-alpha",
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset_file  = \"dataset.csv\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "alone-broad",
   "metadata": {},
   "outputs": [],
   "source": [
    "seed_number = 122"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "adjacent-phoenix",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "data = pd.read_csv(\"dataset.csv\", usecols = ['author','text'])\n",
    "#data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "covered-literacy",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(25000, 2)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "basic-nickel",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.naive_bayes import MultinomialNB\n",
    "\n",
    "# pip install xgboost\n",
    "# xgboost-1.5.2 used\n",
    "from xgboost import XGBClassifier\n",
    "\n",
    "from sklearn.metrics import accuracy_score,classification_report, confusion_matrix\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.metrics import plot_confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "friendly-projector",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import TfidfVectorizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "flying-jerusalem",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Train, Test, Validation Spit ratios 60%, 20%, 20% \n",
    "\n",
    "X_train, X_val, y_train, y_val = train_test_split(data[\"text\"], data[\"author\"], test_size=0.2, random_state=2000)\n",
    "X_train, X_test, y_train, y_test = train_test_split(X_train, y_train, test_size=0.25, random_state=seed_number) # 0.25 x 0.8 = 0.2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "biological-nevada",
   "metadata": {},
   "outputs": [],
   "source": [
    "# XGB\n",
    "# Ref : https://xgboost.readthedocs.io/en/latest/treemethod.html\n",
    "# clf = Pipeline([('tfidf',TfidfVectorizer()),('clf',XGBClassifier(n_estimators = 100, tree_method='gpu_hist', gpu_id=0))])\n",
    "\n",
    "clf = Pipeline([('tfidf',TfidfVectorizer()),('clf1',XGBClassifier(n_estimators = 100, tree_method='exact'))])\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "mounted-creator",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:44:26] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Pipeline(steps=[('tfidf', TfidfVectorizer()),\n",
       "                ('clf1',\n",
       "                 XGBClassifier(base_score=0.5, booster='gbtree',\n",
       "                               colsample_bylevel=1, colsample_bynode=1,\n",
       "                               colsample_bytree=1, enable_categorical=False,\n",
       "                               gamma=0, gpu_id=-1, importance_type=None,\n",
       "                               interaction_constraints='',\n",
       "                               learning_rate=0.300000012, max_delta_step=0,\n",
       "                               max_depth=6, min_child_weight=1, missing=nan,\n",
       "                               monotone_constraints='()', n_estimators=100,\n",
       "                               n_jobs=12, num_parallel_tree=1,\n",
       "                               objective='multi:softprob', predictor='auto',\n",
       "                               random_state=0, reg_alpha=0, reg_lambda=1,\n",
       "                               scale_pos_weight=None, subsample=1,\n",
       "                               tree_method='exact', validate_parameters=1,\n",
       "                               verbosity=None))])"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dimensional-renaissance",
   "metadata": {},
   "source": [
    "### Testing with \"test\" split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "artificial-bathroom",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\sklearn\\utils\\deprecation.py:87: FutureWarning: Function plot_confusion_matrix is deprecated; Function `plot_confusion_matrix` is deprecated in 1.0 and will be removed in 1.2. Use one of the class methods: ConfusionMatrixDisplay.from_predictions or ConfusionMatrixDisplay.from_estimator.\n",
      "  warnings.warn(msg, category=FutureWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                        precision    recall  f1-score   support\n",
      "\n",
      "       Charles Dickens       0.78      0.72      0.75      1006\n",
      "          George Eliot       0.81      0.72      0.76       978\n",
      "           Jane Austen       0.78      0.86      0.82      1026\n",
      "           Jules Verne       0.84      0.86      0.85       994\n",
      "Sir Arthur Conan Doyle       0.72      0.76      0.74       996\n",
      "\n",
      "              accuracy                           0.79      5000\n",
      "             macro avg       0.79      0.79      0.79      5000\n",
      "          weighted avg       0.79      0.79      0.79      5000\n",
      "\n",
      "accuracy_score: 0.7862\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred = clf.predict(X_test)\n",
    "plot_confusion_matrix(clf, X_test, y_test, normalize='pred', cmap='Blues', xticks_rotation=45) \n",
    "print(classification_report(y_test, y_pred))\n",
    "print( \"accuracy_score: \" + str(accuracy_score(y_test, y_pred)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "adult-junction",
   "metadata": {},
   "source": [
    "### Testing with \"validation\" split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "monthly-majority",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\sklearn\\utils\\deprecation.py:87: FutureWarning: Function plot_confusion_matrix is deprecated; Function `plot_confusion_matrix` is deprecated in 1.0 and will be removed in 1.2. Use one of the class methods: ConfusionMatrixDisplay.from_predictions or ConfusionMatrixDisplay.from_estimator.\n",
      "  warnings.warn(msg, category=FutureWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                        precision    recall  f1-score   support\n",
      "\n",
      "       Charles Dickens       0.74      0.68      0.71       985\n",
      "          George Eliot       0.80      0.75      0.77       986\n",
      "           Jane Austen       0.79      0.85      0.82      1016\n",
      "           Jules Verne       0.87      0.84      0.85      1043\n",
      "Sir Arthur Conan Doyle       0.71      0.78      0.74       970\n",
      "\n",
      "              accuracy                           0.78      5000\n",
      "             macro avg       0.78      0.78      0.78      5000\n",
      "          weighted avg       0.78      0.78      0.78      5000\n",
      "\n",
      "accuracy_score: 0.7802\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred = clf.predict(X_val)\n",
    "plot_confusion_matrix(clf, X_val, y_val, normalize='pred', cmap='Blues', xticks_rotation=45) \n",
    "print(classification_report(y_val, y_pred))\n",
    "print( \"accuracy_score: \" + str(accuracy_score(y_val, y_pred)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "worth-anthony",
   "metadata": {},
   "source": [
    "### Accuracy averaged over 20 random states"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "searching-breeding",
   "metadata": {},
   "outputs": [],
   "source": [
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "referenced-three",
   "metadata": {},
   "outputs": [],
   "source": [
    "iteration_number = 20"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "ordered-michigan",
   "metadata": {},
   "outputs": [],
   "source": [
    "random_states = range(1,21)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "oriented-avenue",
   "metadata": {},
   "outputs": [],
   "source": [
    "accuracy_test = []\n",
    "accuracy_validation = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "outdoor-disease",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:44:42] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:44:56] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:45:10] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:45:24] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:45:39] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:45:54] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:46:08] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:46:22] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:46:36] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:46:50] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:47:05] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:47:19] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:47:34] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:47:49] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:48:03] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:48:17] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:48:32] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:48:46] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:49:01] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "e:\\Miniconda3\\envs\\tf2\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[03:49:16] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    }
   ],
   "source": [
    "for seed_number in random_states:\n",
    "    # Train, Test, Validation Spit ratios 60%, 20%, 20% \n",
    "\n",
    "    X_train, X_val, y_train, y_val = train_test_split(data[\"text\"], data[\"author\"], test_size=0.2, random_state=2000)\n",
    "    X_train, X_test, y_train, y_test = train_test_split(X_train, y_train, test_size=0.25, random_state=seed_number) # 0.25 x 0.8 = 0.2    \n",
    "    \n",
    "    \n",
    "    clf.fit(X_train, y_train)    \n",
    "    \n",
    "    y_pred = clf.predict(X_test)    \n",
    "    accuracy_test.append (accuracy_score(y_test, y_pred))\n",
    "    \n",
    "    y_pred = clf.predict(X_val)    \n",
    "    accuracy_validation.append (accuracy_score(y_val, y_pred))\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "incident-harrison",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0.7792,\n",
       " 0.7828,\n",
       " 0.7852,\n",
       " 0.7854,\n",
       " 0.7794,\n",
       " 0.779,\n",
       " 0.782,\n",
       " 0.7838,\n",
       " 0.774,\n",
       " 0.7808,\n",
       " 0.7782,\n",
       " 0.786,\n",
       " 0.7822,\n",
       " 0.7854,\n",
       " 0.7834,\n",
       " 0.7844,\n",
       " 0.7806,\n",
       " 0.7864,\n",
       " 0.784,\n",
       " 0.7844]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "brief-constitution",
   "metadata": {},
   "outputs": [],
   "source": [
    "final_test_accuracy = 0\n",
    "for accuracy in accuracy_test:\n",
    "    final_test_accuracy += accuracy\n",
    "final_test_accuracy = final_test_accuracy/iteration_number"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "residential-gilbert",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "averaged test accuracy score: 0.78233\n"
     ]
    }
   ],
   "source": [
    "print( \"averaged test accuracy score: \" + str(final_test_accuracy))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "tracked-france",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0.7774,\n",
       " 0.7818,\n",
       " 0.7788,\n",
       " 0.7724,\n",
       " 0.7776,\n",
       " 0.778,\n",
       " 0.7766,\n",
       " 0.7784,\n",
       " 0.7776,\n",
       " 0.7728,\n",
       " 0.7778,\n",
       " 0.7766,\n",
       " 0.7774,\n",
       " 0.7782,\n",
       " 0.7756,\n",
       " 0.7828,\n",
       " 0.7778,\n",
       " 0.7812,\n",
       " 0.7718,\n",
       " 0.7754]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "colored-account",
   "metadata": {},
   "outputs": [],
   "source": [
    "final_validation_accuracy = 0\n",
    "for accuracy in accuracy_validation:\n",
    "    final_validation_accuracy += accuracy\n",
    "final_validation_accuracy = final_validation_accuracy/iteration_number"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "geographic-steps",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "averaged validation accuracy score: 0.7773000000000001\n"
     ]
    }
   ],
   "source": [
    "print( \"averaged validation accuracy score: \" + str(final_validation_accuracy))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.9 (Tensorflow 2.6)",
   "language": "python",
   "name": "tf2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}