badi-predictor

Zürich’s public pools (Badi) get busy. Very busy on a sunny Saturday afternoon. This project predicts how busy, so you can decide whether it’s worth the trip.

The system collects real-time occupancy data, stores it in TimescaleDB, and trains an XGBoost model on historical patterns — time of day, day of week, weather conditions. The FastAPI backend serves predictions on demand.

It’s been collecting data since launch. The longer it runs, the better it gets.

Classic side project: annoying problem, way too much infrastructure.