This project performs sentiment analysis on customer reviews using Natural Language Processing (NLP) and Machine Learning (Naive Bayes) in Python. It classifies reviews as positive or negative, and includes text preprocessing, TF-IDF vectorization, model training, and evaluation.
- Clean and preprocess raw text (HTML, stopwords, stemming)
- Convert reviews into numerical features using TF-IDF
- Train a Multinomial Naive Bayes classifier
- Achieve 85%+ accuracy on real-world data
- Export trained model and vectorizer using
joblib - Use the model via CLI or integrate into a web app
sentiment-analysis/ ├── sentiment_predictor.ipynb # CLI script to predict sentiment ├── sentiment_model_nb.joblib # Saved ML model ├── tfidf_vectorizer.joblib # Saved TF-IDF vectorizer ├── IMDB_Dataset.csv # Input dataset (or any CSV with review + label) ├── README.md # Project documentation
- Python 3.8+
- pandas
- scikit-learn
- nltk
- joblib
- matplotlib / seaborn (optional for visualization)