notebooks/09_rf_tfidf_plus_guardian.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Random Forest Model using TF/IDF\n",
    "\n",
    "This model will integrate stories from the Guardian dataset with the original dataset and retrain the model.  The true stories will be split 50/50 between the original true stories and stories from the Guardian set.  The total number of true stories will be equal to the Fake ones to avoid class imbalance."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Imports and Constants"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:14:56.326607Z",
     "start_time": "2020-10-21T17:14:37.871575Z"
    }
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.feature_extraction.text import CountVectorizer, TfidfTransformer\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.metrics import classification_report, plot_confusion_matrix\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import ast\n",
    "from sklearn.metrics import precision_recall_curve, auc\n",
    "from sklearn.metrics import accuracy_score, f1_score, recall_score\n",
    "from sklearn.preprocessing import LabelBinarizer\n",
    "from nltk import word_tokenize          \n",
    "from nltk.stem import WordNetLemmatizer\n",
    "from nltk.corpus import stopwords\n",
    "import model_helper_functions\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "import pickle\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:14:56.343081Z",
     "start_time": "2020-10-21T17:14:56.329113Z"
    }
   },
   "outputs": [],
   "source": [
    "DATA_PATH = '../data/'\n",
    "PRE_PROCESSED_DATA_FILE_NAME = 'news_dataset_pre_processed.csv'\n",
    "GUARDIAN_DATA_FILE_NAME = 'guardian_news_dataset_cleaned.csv'\n",
    "RANDOM_STATE = 42\n",
    "pd.options.display.max_seq_items = 20\n",
    "pd.options.display.max_rows = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:14:56.349848Z",
     "start_time": "2020-10-21T17:14:56.346153Z"
    }
   },
   "outputs": [],
   "source": [
    "SAVE_MODEL = False\n",
    "\n",
    "MODEL_PATH = '../models/'\n",
    "MODEL_FILE_NAME = 'rf_tfidf_plus_guardian_model.sav'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:15:05.085191Z",
     "start_time": "2020-10-21T17:14:56.352552Z"
    }
   },
   "outputs": [],
   "source": [
    "df = pd.read_csv(DATA_PATH + PRE_PROCESSED_DATA_FILE_NAME, \n",
    "                 low_memory = False, \n",
    "                 usecols = ['title', 'text', 'label', 'clean_text_tokens', 'clean_title_tokens'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:15:48.926348Z",
     "start_time": "2020-10-21T17:15:05.087432Z"
    }
   },
   "outputs": [],
   "source": [
    "df.clean_text_tokens = df.clean_text_tokens.map(ast.literal_eval)\n",
    "df.clean_title_tokens = df.clean_title_tokens.map(ast.literal_eval)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:15:49.001372Z",
     "start_time": "2020-10-21T17:15:48.928472Z"
    }
   },
   "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>title</th>\n",
       "      <th>text</th>\n",
       "      <th>label</th>\n",
       "      <th>clean_text_tokens</th>\n",
       "      <th>clean_title_tokens</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Donald Trump Sends Out Embarrassing New Year’...</td>\n",
       "      <td>Donald Trump just couldn t wish all Americans ...</td>\n",
       "      <td>fake</td>\n",
       "      <td>[donald, trump, just, couldn, wish, all, ameri...</td>\n",
       "      <td>[donald, trump, sends, out, embarrassing, new,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Drunk Bragging Trump Staffer Started Russian ...</td>\n",
       "      <td>House Intelligence Committee Chairman Devin Nu...</td>\n",
       "      <td>fake</td>\n",
       "      <td>[house, intelligence, committee, chairman, dev...</td>\n",
       "      <td>[drunk, bragging, trump, staffer, started, rus...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Sheriff David Clarke Becomes An Internet Joke...</td>\n",
       "      <td>On Friday, it was revealed that former Milwauk...</td>\n",
       "      <td>fake</td>\n",
       "      <td>[on, it, was, revealed, that, former, milwauke...</td>\n",
       "      <td>[sheriff, david, clarke, becomes, an, internet...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Trump Is So Obsessed He Even Has Obama’s Name...</td>\n",
       "      <td>On Christmas day, Donald Trump announced that ...</td>\n",
       "      <td>fake</td>\n",
       "      <td>[on, christmas, day, donald, trump, announced,...</td>\n",
       "      <td>[trump, is, so, obsessed, he, even, has, obama...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Pope Francis Just Called Out Donald Trump Dur...</td>\n",
       "      <td>Pope Francis used his annual Christmas Day mes...</td>\n",
       "      <td>fake</td>\n",
       "      <td>[pope, francis, used, his, annual, christmas, ...</td>\n",
       "      <td>[pope, francis, just, called, out, donald, tru...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                               title  \\\n",
       "0   Donald Trump Sends Out Embarrassing New Year’...   \n",
       "1   Drunk Bragging Trump Staffer Started Russian ...   \n",
       "2   Sheriff David Clarke Becomes An Internet Joke...   \n",
       "3   Trump Is So Obsessed He Even Has Obama’s Name...   \n",
       "4   Pope Francis Just Called Out Donald Trump Dur...   \n",
       "\n",
       "                                                text label  \\\n",
       "0  Donald Trump just couldn t wish all Americans ...  fake   \n",
       "1  House Intelligence Committee Chairman Devin Nu...  fake   \n",
       "2  On Friday, it was revealed that former Milwauk...  fake   \n",
       "3  On Christmas day, Donald Trump announced that ...  fake   \n",
       "4  Pope Francis used his annual Christmas Day mes...  fake   \n",
       "\n",
       "                                   clean_text_tokens  \\\n",
       "0  [donald, trump, just, couldn, wish, all, ameri...   \n",
       "1  [house, intelligence, committee, chairman, dev...   \n",
       "2  [on, it, was, revealed, that, former, milwauke...   \n",
       "3  [on, christmas, day, donald, trump, announced,...   \n",
       "4  [pope, francis, used, his, annual, christmas, ...   \n",
       "\n",
       "                                  clean_title_tokens  \n",
       "0  [donald, trump, sends, out, embarrassing, new,...  \n",
       "1  [drunk, bragging, trump, staffer, started, rus...  \n",
       "2  [sheriff, david, clarke, becomes, an, internet...  \n",
       "3  [trump, is, so, obsessed, he, even, has, obama...  \n",
       "4  [pope, francis, just, called, out, donald, tru...  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:15:52.822539Z",
     "start_time": "2020-10-21T17:15:49.003750Z"
    }
   },
   "outputs": [],
   "source": [
    "df_guard = pd.read_csv(DATA_PATH + GUARDIAN_DATA_FILE_NAME, \n",
    "                 low_memory = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:03.542457Z",
     "start_time": "2020-10-21T17:15:52.827150Z"
    }
   },
   "outputs": [],
   "source": [
    "df_guard.tt_tokens = df_guard.tt_tokens.map(ast.literal_eval)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Integrate Guardian Stories as true"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## split original data by label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:03.558687Z",
     "start_time": "2020-10-21T17:17:03.545469Z"
    }
   },
   "outputs": [],
   "source": [
    "df_fake = df[df.label == 'fake']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:03.573064Z",
     "start_time": "2020-10-21T17:17:03.560965Z"
    }
   },
   "outputs": [],
   "source": [
    "df_true_og = df[df.label == 'true']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:03.581434Z",
     "start_time": "2020-10-21T17:17:03.575598Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8723.5"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(df_fake)/2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Take Random Sample of original true data\n",
    "\n",
    "Equal to half of fake data so to keep the classes balanced"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:03.591932Z",
     "start_time": "2020-10-21T17:17:03.583423Z"
    }
   },
   "outputs": [],
   "source": [
    "df_true_og_sample = df_true_og.sample(8724, random_state = RANDOM_STATE)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Take Random Sample of Guardian data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:03.605112Z",
     "start_time": "2020-10-21T17:17:03.593893Z"
    }
   },
   "outputs": [],
   "source": [
    "df_guard_sample = df_guard.sample(8723, random_state = RANDOM_STATE)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Assign the label true"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:03.612249Z",
     "start_time": "2020-10-21T17:17:03.607363Z"
    }
   },
   "outputs": [],
   "source": [
    "df_guard_sample['label'] = 'true'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Combine datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:03.817845Z",
     "start_time": "2020-10-21T17:17:03.614422Z"
    }
   },
   "outputs": [],
   "source": [
    "df_true_og_sample['tt_tokens'] = df_true_og_sample['clean_title_tokens'] + df_true_og_sample['clean_text_tokens']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:04.967968Z",
     "start_time": "2020-10-21T17:17:03.820024Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/freethrall/anaconda3/envs/learn-env/lib/python3.6/site-packages/ipykernel_launcher.py:1: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  \"\"\"Entry point for launching an IPython kernel.\n"
     ]
    }
   ],
   "source": [
    "df_fake['tt_tokens'] = df_fake['clean_title_tokens'] + df_fake['clean_text_tokens']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:05.027736Z",
     "start_time": "2020-10-21T17:17:04.970763Z"
    }
   },
   "outputs": [],
   "source": [
    "df_all = pd.concat([df_fake[['label', 'tt_tokens']], \n",
    "                         df_true_og_sample[['label', 'tt_tokens']], \n",
    "                        df_guard_sample[['label', 'tt_tokens']]], axis=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Split Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:05.036679Z",
     "start_time": "2020-10-21T17:17:05.030037Z"
    }
   },
   "outputs": [],
   "source": [
    "y = df_all.label\n",
    "X = df_all.drop('label', axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:05.092932Z",
     "start_time": "2020-10-21T17:17:05.039045Z"
    }
   },
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, \n",
    "                                                    y,\n",
    "                                                   test_size = 0.2,\n",
    "                                                   random_state = RANDOM_STATE,\n",
    "                                                   stratify = y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:05.100610Z",
     "start_time": "2020-10-21T17:17:05.095069Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(27915, 1)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:05.112505Z",
     "start_time": "2020-10-21T17:17:05.102841Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(6979, 1)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Encode Labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:05.118080Z",
     "start_time": "2020-10-21T17:17:05.114733Z"
    }
   },
   "outputs": [],
   "source": [
    "le = LabelEncoder()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:05.144079Z",
     "start_time": "2020-10-21T17:17:05.126069Z"
    }
   },
   "outputs": [],
   "source": [
    "y_train_enc = le.fit_transform(y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:05.154391Z",
     "start_time": "2020-10-21T17:17:05.149216Z"
    }
   },
   "outputs": [],
   "source": [
    "y_test_enc = le.transform(y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:05.168627Z",
     "start_time": "2020-10-21T17:17:05.156721Z"
    }
   },
   "outputs": [],
   "source": [
    "def train_and_eval_model(X_train, X_test, y_train_enc, y_test_enc, classes_,\n",
    "                        preprocessor, tokenizer, max_df=1.0, ngram_range = (1, 1),\n",
    "                        min_samples_split=2,\n",
    "                        max_depth=None):\n",
    "    \"\"\"\n",
    "    Train and Evaluate and Bag of Words Representation with a Random Forest\n",
    "    classifier.\n",
    "    \"\"\"\n",
    "    \n",
    "    pipeline = Pipeline([\n",
    "    ('bow',CountVectorizer(min_df = 5, \n",
    "                           preprocessor = preprocessor, \n",
    "                           tokenizer = tokenizer, \n",
    "                           max_df = max_df, \n",
    "                           ngram_range = ngram_range)),\n",
    "    ('tfidf', TfidfTransformer()),\n",
    "    ('classifier', RandomForestClassifier(n_jobs = -1, \n",
    "                                          random_state = RANDOM_STATE, \n",
    "                                          min_samples_split = min_samples_split, \n",
    "                                          max_depth = max_depth)),\n",
    "    ])\n",
    "    \n",
    "    pipeline.fit(X_train,y_train_enc)\n",
    "    \n",
    "    model_helper_functions.confustion_matrix_and_classification_report(pipeline, \n",
    "                                                X_train, \n",
    "                                                y_train_enc, \n",
    "                                                classes_, \n",
    "                                               'Training')\n",
    "\n",
    "    model_helper_functions.confustion_matrix_and_classification_report(pipeline, \n",
    "                                                X_test, \n",
    "                                                y_test_enc, \n",
    "                                                classes_,\n",
    "                                                'Test')\n",
    "    \n",
    "    return pipeline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Title & Text only - lowercase, only stopwords, but a larger set of them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:05.182921Z",
     "start_time": "2020-10-21T17:17:05.172585Z"
    }
   },
   "outputs": [],
   "source": [
    "# using expanded stopwords list from https://gist.github.com/sebleier/554280\n",
    "\n",
    "gist_file = open(DATA_PATH + \"gist_stopwords.txt\", \"r\")\n",
    "try:\n",
    "    content = gist_file.read()\n",
    "    expanded_stopwords = content.split(\",\")\n",
    "finally:\n",
    "    gist_file.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Remove 'via' from the list of stopwords b/c it is particular to this dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:05.191330Z",
     "start_time": "2020-10-21T17:17:05.186608Z"
    }
   },
   "outputs": [],
   "source": [
    "expanded_stopwords.remove('via')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Remove UK specific terms that are affecting the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:05.197319Z",
     "start_time": "2020-10-21T17:17:05.193889Z"
    }
   },
   "outputs": [],
   "source": [
    "expanded_stopwords.remove('eu')\n",
    "expanded_stopwords.remove('uk')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:17:05.211131Z",
     "start_time": "2020-10-21T17:17:05.202578Z"
    }
   },
   "outputs": [],
   "source": [
    "def lowercase_and_only_expanded_stopwords(doc):\n",
    "    \"\"\"Remove stopwords and lowercase tokens\"\"\"\n",
    "    stop_words = expanded_stopwords\n",
    "    return [token.lower() for token in doc if token.lower() in stop_words]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:37:47.768827Z",
     "start_time": "2020-10-21T17:17:05.213577Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification Report for Training Set\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        fake       0.92      0.96      0.94     13957\n",
      "        true       0.96      0.92      0.94     13958\n",
      "\n",
      "    accuracy                           0.94     27915\n",
      "   macro avg       0.94      0.94      0.94     27915\n",
      "weighted avg       0.94      0.94      0.94     27915\n",
      "\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"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification Report for Test Set\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        fake       0.88      0.91      0.89      3490\n",
      "        true       0.90      0.88      0.89      3489\n",
      "\n",
      "    accuracy                           0.89      6979\n",
      "   macro avg       0.89      0.89      0.89      6979\n",
      "weighted avg       0.89      0.89      0.89      6979\n",
      "\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"
    },
    {
     "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": [
    "pipe_expanded_sw_no_via = train_and_eval_model(X_train['tt_tokens'], \n",
    "                                                        X_test['tt_tokens'],\n",
    "                                                        y_train_enc, \n",
    "                                                        y_test_enc, \n",
    "                                                        le.classes_, \n",
    "                                                        lowercase_and_only_expanded_stopwords, \n",
    "                                                        model_helper_functions.passthrough,\n",
    "                                                 min_samples_split = 0.005)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:37:47.892790Z",
     "start_time": "2020-10-21T17:37:47.771547Z"
    },
    "scrolled": true
   },
   "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>importance</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>said</th>\n",
       "      <td>0.124759</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>the</th>\n",
       "      <td>0.040807</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>its</th>\n",
       "      <td>0.040782</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>just</th>\n",
       "      <td>0.040064</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>this</th>\n",
       "      <td>0.032734</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>on</th>\n",
       "      <td>0.030779</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>you</th>\n",
       "      <td>0.029564</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>like</th>\n",
       "      <td>0.024284</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>in</th>\n",
       "      <td>0.023287</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>would</th>\n",
       "      <td>0.019989</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>even</th>\n",
       "      <td>0.017526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>here</th>\n",
       "      <td>0.014972</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>actually</th>\n",
       "      <td>0.014146</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>had</th>\n",
       "      <td>0.013308</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>could</th>\n",
       "      <td>0.012963</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>which</th>\n",
       "      <td>0.012526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>doesn</th>\n",
       "      <td>0.012175</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>him</th>\n",
       "      <td>0.011891</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>didn</th>\n",
       "      <td>0.011878</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>know</th>\n",
       "      <td>0.009831</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>isn</th>\n",
       "      <td>0.009666</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>below</th>\n",
       "      <td>0.008534</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>why</th>\n",
       "      <td>0.008465</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>of</th>\n",
       "      <td>0.008243</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last</th>\n",
       "      <td>0.008170</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>so</th>\n",
       "      <td>0.008146</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>how</th>\n",
       "      <td>0.008106</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>re</th>\n",
       "      <td>0.008083</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>is</th>\n",
       "      <td>0.008048</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>really</th>\n",
       "      <td>0.007412</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          importance\n",
       "said        0.124759\n",
       "the         0.040807\n",
       "its         0.040782\n",
       "just        0.040064\n",
       "this        0.032734\n",
       "on          0.030779\n",
       "you         0.029564\n",
       "like        0.024284\n",
       "in          0.023287\n",
       "would       0.019989\n",
       "even        0.017526\n",
       "here        0.014972\n",
       "actually    0.014146\n",
       "had         0.013308\n",
       "could       0.012963\n",
       "which       0.012526\n",
       "doesn       0.012175\n",
       "him         0.011891\n",
       "didn        0.011878\n",
       "know        0.009831\n",
       "isn         0.009666\n",
       "below       0.008534\n",
       "why         0.008465\n",
       "of          0.008243\n",
       "last        0.008170\n",
       "so          0.008146\n",
       "how         0.008106\n",
       "re          0.008083\n",
       "is          0.008048\n",
       "really      0.007412"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feat_imp_expanded_sw_no_via = pd.DataFrame(pipe_expanded_sw_no_via['classifier'].feature_importances_, \n",
    "                                       index = pipe_expanded_sw_no_via['bow'].get_feature_names(), \n",
    "                                       columns = ['importance']).sort_values('importance', \n",
    "                                                                             ascending=False)\n",
    "feat_imp_expanded_sw_no_via.head(30)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Save Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:37:47.901361Z",
     "start_time": "2020-10-21T17:37:47.895755Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model directory already exists\n"
     ]
    }
   ],
   "source": [
    "try: \n",
    "    os.mkdir(MODEL_PATH)\n",
    "except FileExistsError:\n",
    "    print('Model directory already exists')\n",
    "    pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-10-21T17:37:47.927846Z",
     "start_time": "2020-10-21T17:37:47.906698Z"
    }
   },
   "outputs": [],
   "source": [
    "if SAVE_MODEL:\n",
    "    pickle.dump(pipe_expanded_sw_no_via, open(MODEL_PATH + MODEL_FILE_NAME, 'wb'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Summary"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The accuracy and F1 scores for this model are somewhat worse than the model trained on the original dataset, 0.89 here vs. 0.94, but this model should generalize better because the news stories labeled true came from more than one news source."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Not surprisingly the list of important features changed compared to what the model found for the previous version of the dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "learn-env",
   "language": "python",
   "name": "learn-env"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {
    "height": "calc(100% - 180px)",
    "left": "10px",
    "top": "150px",
    "width": "188.391px"
   },
   "toc_section_display": true,
   "toc_window_display": true
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}