PlanetTerp Predictor
An AI-powered tool that forecasts professor grades and course difficulty using historical data from PlanetTerp, helping students make informed registration decisions.
Overview
This project was built to explore whether student review sentiment could reliably forecast professor ratings and course difficulty. By pulling thousands of reviews from the PlanetTerp API, we trained regression models to correlate review tone with quantitative instructor ratings.
Highlights
Scraped and processed 4,544+ student reviews through an NLP pipeline, hitting an 0.788 R-squared score.
Engineered review text features and tuned Random Forest and Gradient Boosting models.
Extracted sentiment polarity to map how review enthusiasm ties back to numeric evaluation scores.
Results
Student sentiment in written comments turned out to be a solid predictor of numeric ratings, with Gradient Boosting outperforming other models across different academic departments.