Sentiment Analysis API

Overview

A simple sentiment analysis project that predicts whether a given text is positive or negative.

• Trained using Python and scikit-learn

• Uses a Naive Bayes model with pipelines for preprocessing and prediction

• Exposed via a FastAPI backend for programmatic access

Tech Stack

• Python – core language

• scikit-learn – model training and pipelines

• FastAPI – REST API backend

• Joblib – model serialization

Features

• Predict sentiment of text input (positive/negative)

• Modular pipeline for preprocessing and model inference

• Simple API endpoint to integrate with other applications

Getting Started

1. Clone the Repository

git clone https://github.com/Poorna-Raj/simple-sentiment-analysis-model.git
cd sentiment-analysis

2. Install Dependencies

pip install -r requirements.txt

3. Get the dataset and move to data folder

https://www.kaggle.com/datasets/lakshmi25npathi/imdb-dataset-of-50k-movie-reviews

4. Run the API

cd app
fastapi dev main.py

Api Usage

1. Visit the Api documentation(Swagger)

http://127.0.0.1:8000/docs

Model Details

• Algorithm: Multinomial Naive Bayes

• Preprocessing: Tokenization, vectorization using CountVectorizer, handled via scikit-learn pipeline

• Accuracy: ~86% on test set