Sentiment-Analysis-API

A FastAPI-based REST API that predicts sentiment (positive, neutral, negative) from text using a trained Logistic Regression model.

1. Clone/Download the Project
2. Install Dependencies

   pip install -r requirements.txt

3. Prepare Your Model Make sure you have these files in your project directory:

• preprocessed.xlsx - training dataset
• main.py - The FastAPI application
• requirements.txt - Dependencies 4.Train and Save the Model

  python log_model.py

This will create sentiment_model.pkl file.

5.Start the API server in first terminal

python main.py

6.Test the API Open your browser and go to: API Documentation: http://localhost:8000/docs API Usage

   curl -X POST "http://localhost:8000/predict" \
        -H "Content-Type: application/json" \
        -d '{"text": "This is amazing let's proceed"}'

Response:

{
"label": "positive",
"confidence": 0.886,
"probabilities": {
 "negative": 0.05,
 "neutral": 0.08,
 "positive": 0.886
}
}

Project Structure

   sentiment-api
   main.py                # FastAPI application
   log_model.py           # Script to train and save model
   test_api.py            # Test script
   requirements.txt       # Python dependencies
   README.md              # This file
   preprocessed.xlsx      # Training dataset
   sentiment_model.pkl    # Trained model (generated)

Model Parameters -The model uses:

   -Algorithm: Logistic Regression
   -Vectorization: TF-IDF with 1000 features
   -N-grams: Unigrams and bigrams (1,2)
   -Classes: negative (-1), neutral (0), positive (1)

Testing

python test_api.py

While the main.py is running, run this command in the second terminal for testing the model by giving custom input texts

Manual Testing

   1.Start the API server
   2.Go to http://localhost:8000/docs
   3.Try the /predict endpoint with different texts:
       "This is amazing!" (should predict positive)
        "Not interested" (should predict negative)