TSA2_Vectorization.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "2f7516a4",
   "metadata": {},
   "source": [
    "![Twitter Sentiment Analysis](tt.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8c708982-e212-4465-bee0-2cb84b0c63ac",
   "metadata": {},
   "source": [
    "# Sentiment Analysis From Scratch: A Deep Dive into Twitter Data\n",
    "\n",
    "### Notebook 2: Word Vectorization\n",
    "In the vast world of Natural Language Processing (NLP), transforming textual data into a numerical format is a foundational step. This notebook delves deep into the vectorization aspect of the sentiment analysis project, turning raw tweet text into vectors that can be fed into machine learning algorithms.\n",
    "\n",
    "The journey begins with the preparation of the dataset, splitting it into training and testing sets, ensuring a robust evaluation process. Following this, several vectorization techniques are explored, each with its unique approach and advantages:"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "25ff4d6d-d022-4551-9a17-a13b45bd2c1e",
   "metadata": {},
   "source": [
    "**CountVectorizer:** A simple yet effective method that transforms the text into vectors based on word count.\n",
    "\n",
    "**TF-IDF Vectorizer:** Going a step beyond mere counting, this method weighs terms based on their importance in the corpus, distinguishing between common and rare terms.\n",
    "\n",
    "**Word2Vec:** A more advanced technique that captures the semantic meaning of words based on their context. Both Skip-Gram and CBOW models are explored to achieve this"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8f230684-d508-41ee-a0f7-9feed904c8c8",
   "metadata": {},
   "source": [
    "Additionally, a logistic regression model is trained and evaluated as a preliminary step to gauge the effectiveness of the vectorized data. This notebook sets the stage for subsequent modeling and deep analysis, ensuring the data is well-prepared and understood.\n",
    "\n",
    "---\n",
    "\n",
    "## Importing Libs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "e551a5d0",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from sklearn.model_selection import train_test_split\n",
    "from datetime import datetime\n",
    "from gensim.models import Word2Vec\n",
    "from nltk import tokenize\n",
    "import numpy as np\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.feature_extraction.text import CountVectorizer,TfidfVectorizer\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import f1_score, accuracy_score, precision_score, recall_score\n",
    "from transformers import BertTokenizer, BertModel\n",
    "import torch\n",
    "from torch.utils.data import DataLoader, Dataset, RandomSampler, SequentialSampler\n",
    "from tqdm.notebook import tqdm\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import os\n",
    "from tqdm import tqdm\n",
    "from IPython.display import clear_output\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2af946f0",
   "metadata": {},
   "source": [
    "## Reading cleaned dataframe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "5d724194",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>target</th>\n",
       "      <th>text</th>\n",
       "      <th>text2</th>\n",
       "      <th>text3</th>\n",
       "      <th>text4</th>\n",
       "      <th>text5</th>\n",
       "      <th>text6</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>@switchfoot http://twitpic.com/2y1zl - Awww, t...</td>\n",
       "      <td>@switchfoot - awww, that's bummer. shoulda got...</td>\n",
       "      <td>switchfoot awww bummer shoulda got david carr ...</td>\n",
       "      <td>switchfoot awww bummer shoulda got david carr ...</td>\n",
       "      <td>switchfoot awww bummer shoulda got david carr ...</td>\n",
       "      <td>switchfoot awww bummer shoulda got david carr ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>is upset that he can't update his Facebook by ...</td>\n",
       "      <td>upset can't update facebook texting it... migh...</td>\n",
       "      <td>upset update facebook texting ... might cry re...</td>\n",
       "      <td>upset update facebook texting ... might cry re...</td>\n",
       "      <td>upset updat facebook text ... might cri result...</td>\n",
       "      <td>upset updat facebook text might cri result sch...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>@Kenichan I dived many times for the ball. Man...</td>\n",
       "      <td>@kenichan dived many times ball. managed save ...</td>\n",
       "      <td>kenichan dived many times ball managed save 50...</td>\n",
       "      <td>kenichan dived many times ball managed save 50...</td>\n",
       "      <td>kenichan dive mani time ball manag save 50 res...</td>\n",
       "      <td>kenichan dive mani time ball manag save rest g...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   target                                               text  \\\n",
       "0       0  @switchfoot http://twitpic.com/2y1zl - Awww, t...   \n",
       "1       0  is upset that he can't update his Facebook by ...   \n",
       "2       0  @Kenichan I dived many times for the ball. Man...   \n",
       "\n",
       "                                               text2  \\\n",
       "0  @switchfoot - awww, that's bummer. shoulda got...   \n",
       "1  upset can't update facebook texting it... migh...   \n",
       "2  @kenichan dived many times ball. managed save ...   \n",
       "\n",
       "                                               text3  \\\n",
       "0  switchfoot awww bummer shoulda got david carr ...   \n",
       "1  upset update facebook texting ... might cry re...   \n",
       "2  kenichan dived many times ball managed save 50...   \n",
       "\n",
       "                                               text4  \\\n",
       "0  switchfoot awww bummer shoulda got david carr ...   \n",
       "1  upset update facebook texting ... might cry re...   \n",
       "2  kenichan dived many times ball managed save 50...   \n",
       "\n",
       "                                               text5  \\\n",
       "0  switchfoot awww bummer shoulda got david carr ...   \n",
       "1  upset updat facebook text ... might cri result...   \n",
       "2  kenichan dive mani time ball manag save 50 res...   \n",
       "\n",
       "                                               text6  \n",
       "0  switchfoot awww bummer shoulda got david carr ...  \n",
       "1  upset updat facebook text might cri result sch...  \n",
       "2  kenichan dive mani time ball manag save rest g...  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tweets = pd.read_csv('training.1600000.processed.noemoticon_cleaned.csv')\n",
    "tweets.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bb507db6",
   "metadata": {},
   "source": [
    "## Splitting data into train and test\n",
    "\n",
    "Before training any machine learning model, it's essential to split the data into training and testing sets. This allows the model to be trained on a portion of the data and tested on an unseen part, ensuring a fair evaluation of performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "c570b7d4",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(tweets['text6'], tweets['target'], random_state = 5, train_size=0.8)\n",
    "X_train = X_train.reset_index(drop=True)\n",
    "y_train = y_train.reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "300f2838",
   "metadata": {},
   "source": [
    "## Logistic Regression: Functions for Model Training and Evaluation\n",
    "\n",
    "Logistic regression offers a straightforward and effective approach to binary classification. In this section, we define functions to train the model and evaluate its performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "40177065",
   "metadata": {},
   "outputs": [],
   "source": [
    "def classificate_lr (X_train, X_test, y_train, y_test):\n",
    "    t1 = datetime.now()\n",
    "    logistic_regression = LogisticRegression(solver = \"lbfgs\", max_iter=10000)\n",
    "    logistic_regression.fit(X_train, y_train)\n",
    "    t2 = datetime.now()\n",
    "    \n",
    "    y_pred = logistic_regression.predict(X_test)\n",
    "    \n",
    "    print('Trained model. The training process lasted {} seconds.'.format((t2-t1).total_seconds(), '2f'))\n",
    "    \n",
    "    accuracy = metrics.accuracy_score(y_test, y_pred)\n",
    "    precision = metrics.precision_score(y_test, y_pred)\n",
    "    recall = metrics.recall_score(y_test, y_pred)\n",
    "    f1 = metrics.f1_score(y_test, y_pred)\n",
    "          \n",
    "    print('Accuracy Score: {:.2f}%'.format(accuracy * 100))\n",
    "    print('Precision Score: {:.2f}%'.format(precision * 100))\n",
    "    print('Recall Score: {:.2f}%'.format(recall * 100))\n",
    "    print('F1 Score: {:.2f}%'.format(f1 * 100))\n",
    "    \n",
    "    return logistic_regression, y_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "428ddb63",
   "metadata": {},
   "outputs": [],
   "source": [
    "def conf_matrix(y_test, y_pred):\n",
    "    cm = metrics.confusion_matrix(y_test, y_pred)\n",
    "    plt.figure(figsize=(10, 10))\n",
    "    ax= plt.subplot()\n",
    "    hm = sns.heatmap(cm, annot=True, fmt='g', ax=ax, cmap=\"Blues\", square=True);  #annot=True to annotate cells, ftm='g' to disable scientific notation\n",
    "    \n",
    "    ax.set_xlabel('Predicted label');\n",
    "    ax.set_ylabel('True label'); \n",
    "    ax.set_title('Confusion Matrix'); \n",
    "    ax.xaxis.set_ticklabels(['0', '1']); \n",
    "    ax.yaxis.set_ticklabels(['0', '1']);   "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fc56749b",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "## Exploring Vectorization Methods\n",
    "\n",
    "In Natural Language Processing (NLP), raw textual data is inherently unstructured and cannot be directly fed into machine learning algorithms, which require numerical input. Vectorization bridges this gap by converting text into numerical vectors, representing the content and structure of the original data in a format that algorithms can understand and process. By doing so, vectorization unlocks the potential to derive meaningful insights and patterns from textual data.\n",
    "\n",
    "\n",
    "### CountVectorizer\n",
    "This is a simple method for vectorizing text data, part of the Bag of Words (BoW) approach. It transforms the text into vectors by counting the number of times each word appears in the text. This results in a sparse matrix where each row corresponds to a document and each column is a unique word from the entire corpus"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "f78cbf76",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trained model. The training process lasted 4.282345 seconds.\n",
      "Accuracy Score: 59.50%\n",
      "Precision Score: 58.63%\n",
      "Recall Score: 64.72%\n",
      "F1 Score: 61.52%\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "count_vectorizer = CountVectorizer(max_features=300)\n",
    "X_train_cv = count_vectorizer.fit_transform(X_train)\n",
    "X_test_cv = count_vectorizer.fit_transform(X_test)\n",
    "\n",
    "lr_cv, y_pred_cv = classificate_lr(X_train_cv, X_test_cv, y_train, y_test)\n",
    "conf_matrix(y_test, y_pred_cv)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5914be32",
   "metadata": {},
   "source": [
    "### TF-IDF Vectorizer\n",
    "This method also transforms text into vectors, but it does so using the Term Frequency-Inverse Document Frequency (TF-IDF) weighting scheme. It calculates the frequency of each word in a document (Term Frequency), but then scales these counts based on how common the word is across all documents (Inverse Document Frequency). This helps to highlight words that are more unique to a document, giving them more weight than common words."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "ef9dc18e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trained model. The training process lasted 5.114 seconds.\n",
      "Accuracy Score: 59.41%\n",
      "Precision Score: 58.70%\n",
      "Recall Score: 63.63%\n",
      "F1 Score: 61.07%\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "tfidf_vectorizer = TfidfVectorizer(max_features=300)\n",
    "\n",
    "X_train_tfidf = tfidf_vectorizer.fit_transform(X_train)\n",
    "X_test_tfidf = tfidf_vectorizer.fit_transform(X_test)\n",
    "\n",
    "lr_tfidf, y_pred_tfidf = classificate_lr(X_train_tfidf, X_test_tfidf, y_train, y_test)\n",
    "conf_matrix(y_test, y_pred_tfidf)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "72dea7bb",
   "metadata": {},
   "source": [
    "### Vectorization with Word2Vec: Training Skip-Gram and CBOW Models\n",
    "\n",
    "Word2Vec is a groundbreaking technique in the realm of Natural Language Processing (NLP) that seeks to represent words as dense vectors in a continuous vector space. Unlike traditional methods that consider words in isolation, Word2Vec captures semantic relationships between words based on their context. This context-aware representation allows for capturing nuances like synonyms, antonyms, and even analogies.\n",
    "\n",
    "There are two main architectures in Word2Vec:"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5f2d571a-f052-4d06-99a1-4195deaa37f2",
   "metadata": {},
   "source": [
    "<b>Skip-Gram</b>: This model predicts the context (surrounding words) given a target word. It works well with a small amount of data and captures rare words or phrases effectively.\r",
    "<b>\n",
    "CBOW (Continuous Bag of Words)</b>: This model predicts the target word from its context. It's faster and has better representations for more frequent words compared to Skip-Gram\n",
    "\n",
    "By training Word2Vec models, words with similar meanings tend to cluster together in the vector space, allowing for more accurate and nuanced natural language processing tasks.."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "9c819e7f",
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_w2v(dataset, skipgram=0, w=2, s=300, mc=25, a=0.01, ma=0.005):\n",
    "    w2v_model = Word2Vec(sg=skipgram, window=w, vector_size=s, min_count=mc, alpha=a, min_alpha=ma)\n",
    "    lst_tokens = [str(text).split(' ') for text in dataset]\n",
    "    w2v_model.build_vocab(lst_tokens)\n",
    "    w2v_model.train(lst_tokens, total_examples=w2v_model.corpus_count,epochs=30)\n",
    "    return w2v_model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "edeb2021",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trained cbow model. The training process lasted 211.836448 seconds.\n"
     ]
    }
   ],
   "source": [
    "t1 = datetime.now()\n",
    "w2v_cbow = train_w2v(X_train)\n",
    "t2 = datetime.now()\n",
    "\n",
    "print('Trained cbow model. The training process lasted {} seconds.'.format((t2-t1).total_seconds(), '2f'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "ec3b2e7a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trained skipgram model. The training process lasted 373.069168 seconds.\n"
     ]
    }
   ],
   "source": [
    "t1 = datetime.now()\n",
    "w2v_sg = train_w2v(X_train, 1)\n",
    "t2 = datetime.now()\n",
    "\n",
    "print('Trained skipgram model. The training process lasted {} seconds.'.format((t2-t1).total_seconds(), '2f'))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c350aebb",
   "metadata": {},
   "source": [
    "### Testing cbow and skipgram models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "dcab133f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('luv', 0.5932999849319458),\n",
       " ('looov', 0.45908457040786743),\n",
       " ('amaz', 0.45163121819496155),\n",
       " ('loooov', 0.44123411178588867),\n",
       " ('awesom', 0.40450987219810486),\n",
       " ('lovin', 0.3972564935684204),\n",
       " ('loov', 0.39662566781044006),\n",
       " ('ador', 0.3920896649360657),\n",
       " ('loveeee', 0.3900236487388611),\n",
       " ('lovee', 0.3682140111923218)]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "w2v_cbow.wv.most_similar('love')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "d73995dc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('stupid', 0.4828619658946991),\n",
       " ('suck', 0.45392540097236633),\n",
       " ('dislik', 0.4536099135875702),\n",
       " ('ugh', 0.3873719573020935),\n",
       " ('despis', 0.3677579164505005),\n",
       " ('curs', 0.3641287386417389),\n",
       " ('ughh', 0.3634510636329651),\n",
       " ('horribl', 0.3626236617565155),\n",
       " ('fml', 0.3562828302383423),\n",
       " ('grrrrr', 0.35588908195495605)]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "w2v_cbow.wv.most_similar('hate')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "e7f9e323",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('loveeee', 0.557395339012146),\n",
       " ('luv', 0.5281370878219604),\n",
       " ('loov', 0.49940598011016846),\n",
       " ('loooov', 0.4886651039123535),\n",
       " ('lovee', 0.4662560522556305),\n",
       " ('looooooov', 0.4335663616657257),\n",
       " ('looooov', 0.4304094612598419),\n",
       " ('looov', 0.42621925473213196),\n",
       " ('godbless', 0.4245189428329468),\n",
       " ('lovin', 0.41528207063674927)]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "w2v_sg.wv.most_similar('love')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "844597d2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('suck', 0.46546056866645813),\n",
       " ('detest', 0.4313935935497284),\n",
       " ('dislik', 0.42485353350639343),\n",
       " ('despis', 0.42364123463630676),\n",
       " ('stupid', 0.415779709815979),\n",
       " ('goddamnit', 0.3843092918395996),\n",
       " ('suckss', 0.38224416971206665),\n",
       " ('bane', 0.37656620144844055),\n",
       " ('dictat', 0.37384697794914246),\n",
       " ('ugh', 0.3725864589214325)]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "w2v_sg.wv.most_similar('hate')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "55416f4d",
   "metadata": {},
   "source": [
    "### Vectorizing twitter data with Word2Vec models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "b7ee49a7",
   "metadata": {},
   "outputs": [],
   "source": [
    "ws_tokenizer = tokenize.WhitespaceTokenizer()\n",
    "\n",
    "def vectorizing (text, modelo):\n",
    "    x = len(text)\n",
    "    y = 300\n",
    "    matrix = np.zeros((x,y))\n",
    "    \n",
    "    for i in range(x):\n",
    "        words = ws_tokenizer.tokenize(text.iloc[i])\n",
    "        \n",
    "        for word in words:\n",
    "            if word in modelo.wv:\n",
    "                matrix[i] += modelo.wv.get_vector(word)\n",
    "    return matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "eba5720d",
   "metadata": {},
   "outputs": [],
   "source": [
    "matrix_train_cbow = vectorizing(X_train, w2v_cbow)\n",
    "matrix_test_cbow = vectorizing(X_test, w2v_cbow)\n",
    "\n",
    "matrix_train_sg= vectorizing(X_train, w2v_sg)\n",
    "matrix_test_sg = vectorizing(X_test, w2v_sg)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "688b883d",
   "metadata": {},
   "source": [
    "### Vectorization with Word2Vec: Evaluation of Skip-Gram and CBOW Models\n",
    "\n",
    "#### CBOW (Continuous Bag of Words)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "61feda65",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trained model. The training process lasted 36.984022 seconds.\n",
      "Accuracy Score: 75.41%\n",
      "Precision Score: 74.41%\n",
      "Recall Score: 77.50%\n",
      "F1 Score: 75.92%\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#word2vec cbow model\n",
    "lr_cbow, y_pred_cbow = classificate_lr(matrix_train_cbow, matrix_test_cbow, y_train, y_test)\n",
    "conf_matrix(y_test, y_pred_cbow)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8857f56a",
   "metadata": {},
   "source": [
    "#### Skip-Gram"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "ddf92142",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trained model. The training process lasted 17.586995 seconds.\n",
      "Accuracy Score: 75.25%\n",
      "Precision Score: 74.42%\n",
      "Recall Score: 77.01%\n",
      "F1 Score: 75.69%\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#word2vec skipgram model\n",
    "lr_sg, y_pred_sg = classificate_lr(matrix_train_sg, matrix_test_sg, y_train, y_test)\n",
    "conf_matrix(y_test, y_pred_sg)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5b6e5223",
   "metadata": {},
   "source": [
    "### Exporting Word2Vec models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "505d5081",
   "metadata": {},
   "outputs": [],
   "source": [
    "w2v_cbow.save(\"models/word2vec_cbow.model\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "01ac4754",
   "metadata": {},
   "outputs": [],
   "source": [
    "w2v_sg.save(\"models/word2vec_sg.model\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e6929623",
   "metadata": {},
   "source": [
    "## Comparative Performance Analysis of Vectorization Methods\r\n",
    "In the task of classifying tweet sentiments, different vectorization methods showcased varying degrees of effectiveness:\r",
    "<b>\n",
    "\r\n",
    "CountVectorizer and TF-IDF Vectoriz</b>er: Both methods exhibited comparable outcomes, with accuracy levels hovering around 59%, precision at approximately 58%, and F1 scores near 61%. Notably, the training phase for these techniques was swift, concluding in just about 4 to 5 seconds<b>.\r\n",
    "\r\n",
    "Word2Vec (CBOW and Skip-Gram): These models clearly outperformed the previous methods. They achieved accuracy levels exceeding 75%, precision rates surpassing 74%, and F1 scores above 75%. Such results underscore their enhanced capability in discerning the sentiment behind tweets.\r\n",
    "\r\n",
    "In summation, while the Word2Vec models demanded a more extended training period, the trade-off in performance was undoubtedly favorable. Their intricate design, which captures the contextual relationships between words, proves to be advantageous in sentiment analysis endeavors, especially when nuanced understanding is paramount."
   ]
  }
 ],
 "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": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}