Sampling car-sharing data to evaluate urban traffic behaviour

Nowadays, in modern countries, urban traffic per se must be addressed as a problem. The “congestion factor” is fought by introducing regulations to reduce private traffic: tolls, dedicated lanes, narrow lanes, low speed limits, reduced parking availability, etc. Some help can come from car-sharing, i.e., pools of shared vehicles to be rented for short periods of time. Car-sharing vendors “publish” (not entirely/easily accessible) data about the state of their vehicle pool... Can this data be used to analyse the overall traffic behaviour in town? The authors scraped car-sharing vendors’ websites for a couple of years, made data uniform and then queried and graphed the dataset. Some interesting findings are: the “lung effect” (morning moving-in, evening moving-out); evening peak usage (people using car-sharing instead of taxicabs to go out at night for leisure); vehicle usage (the total number of “busy” vehicles at any time) never goes beyond 70%. Moreover, by combining information about parking locations, movement vectors can be drawn to evaluate frequent paths.