🎯 AI-Based-Review-Classifier

📌 Overview

The AI-Based Review Classifier is a Python-based machine learning application that classifies customer reviews as Positive or Negative. It uses a Logistic Regression model trained on TF-IDF features, and exposes a RESTful API for real-time predictions using Flask (with an example FastAPI implementation also included).

---

🚀 Features

• ✅ Logistic Regression Model for Sentiment Analysis
• ✅ TF-IDF Text Vectorization
• ✅ RESTful API for Real-Time Classification (Flask)
• ✅ Preprocessing, Training, and Evaluation Scripts
• ✅ Jupyter Notebooks for Data Exploration and Testing
• ✅ Docker Support for Easy Deployment
• ✅ Automated Model Testing and Evaluation

---

📂 Project Structure

AI-Based-Review-Classifier/
│   ai_service.py                # Python client for API
│   check_data_balance.py        # Data balance checker
│   Dockerfile                   # Docker deployment
│   evaluate.py                  # Model evaluation script
│   main.py                      # FastAPI example (optional)
│   model.pkl                    # Trained model
│   ModelTesting.py              # Model testing script
│   profile_training.py          # Profiling training time
│   sample.py                    # Sample model training
│   vectorizer.pkl               # Saved TF-IDF vectorizer
│   readme.md                    # Project documentation
│
├── api/
│   ├── app.py                   # Flask API for predictions
│   ├── model_loader.py          # Model/vectorizer loader
│   └── requirements.txt         # Python dependencies
│
├── notebooks/
│   ├── DataExploration.ipynb    # Data exploration
│   └── ModelTesting.ipynb       # Model evaluation
│
├── tests/
│   ├── test_predictions.py      # API prediction tests
│   └── test_preprocessing.py    # Preprocessing tests
│
└── training/
    ├── preprocess.py            # Data preprocessing
    ├── train_model.py           # Model training
    └── evaluate.py              # Model evaluation

---

🛠️ Tech Stack

• Python 3.10+
• Flask (REST API)
• Scikit-Learn (Logistic Regression, TF-IDF)
• Pandas, NumPy
• Jupyter Notebooks (Experiments)
• Docker (Deployment)

---

🤖 Model & Algorithm Details

• Algorithm: Logistic Regression (from scikit-learn)
• Feature Extraction: TF-IDF (Term Frequency-Inverse Document Frequency)
• Input: Raw review text
• Output: Sentiment label (positive or negative) and confidence score
• Model Artifacts: model.pkl (Logistic Regression), vectorizer.pkl (TF-IDF)

---

📡 API Endpoints

Flask API

Method	Endpoint	Description
POST	/predict	Classifies review text

Sample Request:

POST http://localhost:5000/predict
Content-Type: application/json
{
  "text": "This product is amazing!"
}

Sample Response:

{
  "prediction": "positive",
  "confidence": "98.00%"
}

---

⚙️ Installation & Setup

1️⃣ Clone the Repository

git clone https://github.com/yourusername/AI-Based-Review-Classifier.git
cd AI-Based-Review-Classifier

2️⃣ Install Dependencies

pip install -r api/requirements.txt

---

🚀 Running the Application

Start the Flask API

cd api
python app.py

The API will be available at http://localhost:5000/predict.

---

🐳 Docker Deployment

To build and run the app with Docker:

docker build -t review-classifier .
docker run -p 8001:8001 review-classifier

---

🧪 Testing

• Run tests/test_predictions.py to test API predictions.
• Run tests/test_preprocessing.py to test preprocessing.
• Use Jupyter notebooks in notebooks/ for exploration and evaluation.

---

🤝 Contributing

We welcome contributions! Please fork the repo, create a branch, and submit a pull request.

---

📜 License

This project is licensed under the MIT License.

🚀 Happy Coding! 🎉