Student Feedback Analyzer

Course feedback gets collected and then sits there. Two hundred free-text comments per subject is more than anyone reads, so what actually happens is someone skims twenty and forms an impression.

A Streamlit app where students submit name, subject, rating and comments; everything lands in MySQL; and the dashboard turns the pile into something readable — average rating per subject as a bar chart, then per subject a word cloud, a Gemini sentiment classification, and a generated summary of every comment at once.

One table, one connection, and the analysis runs on demand rather than on a schedule, so the numbers are always current with the submissions. Credentials and the API key come from a .env file — a lesson carried over from an earlier project.

There is no authentication, so nothing stops one person submitting fifty times. For a class exercise that is fine; for anything with a decision attached to the result it is the first thing that would need fixing.