🛒 Customer Review Sentiment Analyzer

An end-to-end Natural Language Processing (NLP) application that classifies customer reviews into Positive, Neutral, or Negative sentiment using TF-IDF vectorization, Logistic Regression, and an interactive Streamlit interface.

---

Overview

Customer feedback provides valuable insights into product quality and user satisfaction, but manually analyzing large volumes of reviews is time-consuming.

This project presents an end-to-end sentiment analysis pipeline that transforms raw customer reviews into structured sentiment predictions through text preprocessing, TF-IDF feature extraction, machine learning classification, and an interactive Streamlit application.

The application predicts customer sentiment in real time while providing prediction confidence and class probability scores for improved interpretability.

---

Features

• 💬 Real-time customer review sentiment prediction
• 🤖 Logistic Regression classifier
• 📄 TF-IDF text vectorization
• 📊 Prediction confidence scores
• 📈 Class probability visualization
• 🎯 Interactive Streamlit interface
• 📝 Example reviews for quick testing
• 🔁 Reproducible machine learning pipeline

---

Application Workflow

Customer Review
        │
        ▼
Text Preprocessing
        │
        ▼
TF-IDF Vectorization
        │
        ▼
Logistic Regression
        │
        ▼
Sentiment Prediction
        │
        ▼
Confidence Score & Probability Distribution

---

Screenshots

Positive Sentiment Prediction

---

Negative Sentiment Prediction

---

Machine Learning Pipeline

The project follows a traditional supervised learning workflow:

1. Load customer review dataset
2. Split into training and testing sets
3. Convert text into TF-IDF feature vectors
4. Train Logistic Regression classifier
5. Evaluate model performance
6. Save trained model using Joblib
7. Perform real-time inference through Streamlit

---

Technology Stack

Programming Language

• Python

Machine Learning

• Scikit-learn
• Logistic Regression
• TF-IDF Vectorizer

Data Processing

• Pandas
• NumPy

Deployment Interface

• Streamlit

Model Serialization

• Joblib

---

Project Structure

customer_review_sentiment_analyzer/

├── app.py                  # Streamlit application
├── train.py                # Model training script
├── model.pkl               # Trained classifier
├── vectorizer.pkl          # TF-IDF vectorizer
├── requirements.txt
├── README.md
│
├── dataset/
│     └── customer_reviews.csv
│
└── screenshots/
      ├── homepage.png
      ├── positive.png
      └── negative.png

---

Running the Project

Clone the repository

git clone https://github.com/panicAtTheCompile/customer_review_sentiment_analyzer.git

Install dependencies

pip install -r requirements.txt

Launch the application

streamlit run app.py

---

Future Improvements

• Transformer-based sentiment classification (BERT)
• Explainable AI using SHAP
• Cloud deployment
• Batch sentiment analysis
• REST API integration
• Aspect-based sentiment analysis
• Interactive analytics dashboard

---

Author

Harshita Pulavarti

Engineering Undergraduate • IIT Kharagpur

Interested in Artificial Intelligence • Machine Learning • NLP • Data Science

---

License

This project is released under the MIT License.