Natural Language Processing

I developed a product review classification system using machine learning to analyze code-mixed English and Roman Urdu reviews from the Daraz Code Mixed Product Reviews dataset. It integrates Logistic Regression, Support Vector Machines, and Recurrent Neural Networks to predict sentiments (positive, negative, neutral). The Flask-based backend processes inputs, while the React.js front end provides real-time analysis. The system was designed to handle noisy, informal language patterns often seen in online reviews. It demonstrates the trade-offs between classical ML methods and deep learning approaches for text analysis. Overall, the project provided a complete end-to-end pipeline from model training to deployment for real-world usability.