About Sentiment - Analysis

A portfolio-ready sentiment analysis platform built to demonstrate a complete ML workflow — from data preprocessing to a polished production UI.

Dataset

The model is trained on the IMDb Movie Reviews Dataset — a balanced corpus of 50,000 reviews split evenly between positive and negative classes. The dataset is a long-standing benchmark for binary text classification.

Reviews
50,000
Classes
Positive · Negative
Test Accuracy
89%+

Machine Learning Workflow

1. Preprocessing

Lowercase conversion, punctuation removal, stopword filtering, and tokenization.

2. TF-IDF Vectorization

Convert text into weighted numerical features that emphasize meaningful words.

3. Logistic Regression

Train a linear classifier to predict Positive / Negative sentiment.

4. Inference

Serve predictions via a REST API and surface them in this UI with confidence scores.

Technologies

Python
Backend language for the ML training pipeline
Scikit-learn
TF-IDF vectorizer + Logistic Regression model
IMDb Dataset
50,000 labeled movie reviews (25k train / 25k test)
React + TypeScript
Type-safe modern frontend
TanStack Start
Full-stack React framework with SSR
Tailwind CSS
Utility-first styling with custom design tokens

API Reference

The frontend is wired to call this endpoint shape:

POST /predict
Request:
{ "review": "The movie was amazing" }

Response:
{ "sentiment": "Positive", "confidence": 94.5 }