Sentiment Analysis of Product Reviews using Machine Learning

Inspiration

The initial part of this project was a collaborative work done as part of my CSE437 (DATA SCIENCE: CODING WITH REAL WORLD DATA) course.

What it does

Analyze customer product reviews to determine sentiment polarity (positive/negative). Uses ensemble learning with multiple models and TF-IDF text feature extraction for accurate real-time predictions on product reviews.

How I built it

We used a dataset consisting of 40,000 customer reviews from Amazon to train 5 models in total: Decision Tree, Logistic Regression, Support Vector Machine (SVM), and AdaBoost, all on a single Jupyter notebook file. Later on I implemented the prediction pipeline on a webpage using a Flask REST API as the backend ML service, connected to an Express (Node.js) server acting as a proxy, and a React frontend where users can input any review text and get a live sentiment prediction.

Challenges I ran into

Accomplishments that I'm proud of

What I learned

What's next for Sentiment Analysis of Product Reviews using Machine Learning

Built With

• express.js
• flask
• python
• scikit-learn