Currently, the station granularity function just randomly infers riders' locations around a station, ensuring they are nearest the station they'll be using, but otherwise random. Certainly with points-of-interest data (businesses and their operating/busy hours, residences, offices, etc.) a model could be created which would infer locations with more accuracy. This issue has such depth I believe it could almost merit it's own project (meaning, a model which could estimate where a traveller was likely headed to based on where they got off public transit, or parked their car); but it would have excellent applications in this project as well.
Currently, the station granularity function just randomly infers riders' locations around a station, ensuring they are nearest the station they'll be using, but otherwise random. Certainly with points-of-interest data (businesses and their operating/busy hours, residences, offices, etc.) a model could be created which would infer locations with more accuracy. This issue has such depth I believe it could almost merit it's own project (meaning, a model which could estimate where a traveller was likely headed to based on where they got off public transit, or parked their car); but it would have excellent applications in this project as well.