🎬 Movie Sentiment Analysis

Overview

This project performs sentiment analysis on a large dataset of movie reviews using machine learning and natural language processing techniques. The model classifies reviews as positive or negative based on their text content.

Features

• Reads and analyzes a clean CSV dataset (movie.csv) of movie reviews.
• Exploratory Data Analysis (EDA) with visualizations:
  • Distribution of sentiment labels
  • Review length histogram
  • Word count histogram
• Text preprocessing and feature extraction using TF-IDF Vectorization.
• Machine learning model training using Multinomial Naive Bayes.
• Model evaluation with accuracy and classification report.
• Predicts sentiment for new, unseen reviews.

Performance Metrics

Classification Report:

              precision    recall  f1-score   support

           0       0.84      0.89      0.87      5994
           1       0.88      0.83      0.86      6006

    accuracy                           0.86     12000
   macro avg       0.86      0.86      0.86     12000
weighted avg       0.86      0.86      0.86     12000

• Accuracy: 0.86
• Precision (Negative): 0.84
• Recall (Negative): 0.89
• F1-score (Negative): 0.87
• Precision (Positive): 0.88
• Recall (Positive): 0.83
• F1-score (Positive): 0.86

Usage

1. Place your movie.csv file in the same directory as moviesentimentanalysis.py.
2. Run the script:

   python moviesentimentanalysis.py

3. The script will:
   • Print the dataset and column info
   • Show EDA plots (label distribution, review length, word count)
   • Train and evaluate the sentiment analysis model
   • Print accuracy and classification report
   • Predict sentiment for sample new reviews

Example Output

==================================================
🎬 Sentiment Analysis of Movie Reviews
==================================================

Dataset:
   ... (prints first few rows of movie.csv) ...

Sentiment analysis model trained!

Accuracy: 0.86

Classification Report:
              precision    recall  f1-score   support
           0       0.84      0.89      0.87      5994
           1       0.88      0.83      0.86      6006
    accuracy                           0.86     12000
   macro avg       0.86      0.86      0.86     12000
weighted avg       0.86      0.86      0.86     12000

--- Testing with new reviews ---
Review: "The story was incredible and the visuals were stunning." -> Sentiment: Positive
Review: "It was a very dull and slow-paced film." -> Sentiment: Negative
Review: "A decent movie, not great but not bad either." -> Sentiment: Negative

Dependencies

• numpy
• pandas
• scikit-learn
• seaborn
• matplotlib

Install with:

pip install numpy pandas scikit-learn seaborn matplotlib

Acknowledgments

Special thanks to Srihari Sir (IIT Guwahati Faculty) for providing the foundational code, educational guidance, and opportunities to learn advanced machine learning concepts.
Thanks also to Masai School for supporting this learning journey.

---

Most of the code and structure was created by Srihari Sir, with educational spaces and comments for teaching. My contribution was primarily in learning, understanding, and completing the exercises provided.

---