CityPulse has four sources: (a) air-quality sensors (hourly PM2.5 and temperature per sensor), (b) bike-share trips and live station status, (c) monthly energy use of 120 public buildings, (d) bus arrival predictions every 30 seconds.
For each source: name the dataset type(s) (table, network, field, geometry; static or dynamic), the items and keys, and classify every attribute. Then design the tidy tables (names, columns, keys) and the joins needed to answer: "Do districts with worse air quality have fewer bike trips?" Point out two places where an AI tool inferring types from the raw files would probably go wrong.