<<<<<<< HEAD
# NLP Project – Yelp Review Sentiment Analysis
## Overview
This project performs multi-class sentiment analysis on Yelp restaurant reviews using Transformer-based models. The goal is to classify reviews into five sentiment categories corresponding to star ratings (1–5).
## Dataset
- Yelp Review Full Dataset (Hugging Face)
- ~650k training samples, ~50k test samples
- Labels: 1-star to 5-star
## Model
- RoBERTa-base (fine-tuned)
- Multi-class classification (5 classes)
## 🔧 Workflow
1. Data Exploration
2. Text Preprocessing
3. Model Training (with subsampling due to CPU constraints)
4. Model Evaluation (Precision, Recall, F1-score, Confusion Matrix)
## Results
- Accuracy: ~72%
- Weighted F1-score: ~0.62
- Strong performance on extreme sentiment classes
## Project Structure
154aef4 (Initial commit - NLP Yelp Sentiment Analysis)