/NLP-Sentiments

Primary LanguageJupyter Notebook

Sentiment Analysis on IMDB Dataset

Overview

This project is a sentiment analysis task performed on the IMDB dataset. The notebook applies several machine learning models like Logistic Regression, SVM, and Random Forest to predict whether a review is positive or negative. The analysis involves text preprocessing, vectorization techniques (CountVectorizer and TF-IDF), and evaluation using confusion matrices and classification reports.

Table of Contents

Prerequisites

Make sure you have the following libraries installed:

  • Python 3.10+
  • pandas
  • scikit-learn
  • spacy
  • nltk
  • seaborn
  • matplotlib

Required Commands:

  • Install Spacy model: 
    !pip install spacy
!python -m spacy download en_core_web_sm
  • Install NLTK: 
    !pip install nltk

Installation

  1. Clone the repository:

    git clone https://github.com/your-repo/NLPL_Assignment1.git

  2. Open the notebook in Google Colab or Jupyter Notebook:

    NLPL_Assignment_1_095.ipynb

  3. Install the required packages (already in Google Colab):

    !pip install -r requirements.txt

Dataset

The dataset used in this project is the IMDB movie reviews dataset. It contains 50,000 labeled reviews, with an even distribution of positive and negative sentiments.

  • Columns:
    • review: The actual movie review text.
    • sentiment: The label (positive or negative).

The dataset can be downloaded from IMDB Dataset.

Running the Notebook

  1. Load the dataset:

    df = pd.read_csv('/content/drive/MyDrive/IMDB Dataset.csv')

  2. Preprocess the text data using Spacy:

    def clean_text(text):
    text = re.sub(r'http\S+|www\S+|https\S+', '', text, flags=re.MULTILINE)
    text = re.sub(r'[^A-Za-z\s]', '', text)
    doc = nlp(text.lower())
    tokens = [token.lemma_ for token in doc if not token.is_stop]
    return ' '.join(tokens)
df['cleaned_review'] = df['review'].apply(clean_text)

  3. Split the dataset into training and testing sets:

    X_train, X_test, y_train, y_test = train_test_split(df['cleaned_review'], df['sentiment'], test_size=0.2, random_state=42)

  4. Apply vectorization (CountVectorizer and TF-IDF):

    count_vectorizer = CountVectorizer()
tfidf_vectorizer = TfidfVectorizer()

  5. Train and evaluate models using the evaluate_model function:

    def evaluate_model(model, X_train, X_test, y_train, y_test, vectorizer_name):
    model.fit(X_train, y_train)
    y_pred = model.predict(X_test)
    print(classification_report(y_test, y_pred))

Models Used

The following models are implemented:

  • Logistic Regression: A simple linear model for classification.
  • SVM (Support Vector Machine): A robust classifier for binary sentiment classification.
  • Random Forest Classifier: An ensemble model based on decision trees.

Evaluation

The models were evaluated based on the following metrics:

  • Confusion Matrix: Visualized using Seaborn to compare predicted vs. actual results.
  • Classification Report: Precision, recall, f1-score, and accuracy for both positive and negative classes.
  • Model Comparison: Bar plots comparing precision, recall, f1-score, and accuracy across different models.

Results

  • Logistic Regression (CountVectorizer):

    • Accuracy: 88%
    • Precision: 87%
    • Recall: 89%

  • SVC (TF-IDF):

    • Accuracy: 90%
    • Precision: 88%
    • Recall: 91%

  • Random Forest Classifier (TF-IDF):

    • Accuracy: 85%
    • Precision: 85%
    • Recall: 85%

Conclusion

Among the models tested, the SVM model using TF-IDF Vectorizer yielded the best results with an accuracy of 90%, followed closely by Logistic Regression. Random Forest underperformed in comparison.

License

This project is licensed under the MIT License - see the LICENSE file for details.