Analyzing Carsharing “Public” (Scraped) Data to Study Urban Traffic Patterns

During the second half of the twentieth century pollution became a relevant problem, but after the seventies many governments began legislating against emissions. From then on, air pollution decreased as new technologies replaced older ones and now transportation, both private and public, is no longer the main source of pollution in modern countries (it still is in third world areas). Today urban trafficper se is the main problem. The pollution traffic component being relegated in the background, many governments and local administrations address the “congestion factor” by introducing regulations to reduce private traffic, considered the main source of congestion: access tolls and (public transport/car pools) dedicated lanes and odd measures such as narrowing lanes (and/or reducing their number), lowering speed limits, reducing parking availability, etc. Geolocation and road navigation technologies, combined with widespread mobile connectivity infrastructures have enabled researchers to study the evolution of traffic at a great depth. To the extent that some vendor, namely TomTom, uses collected customer navigators’ data to publish annual reports - the “TomTom Traffic Index” - about the state of congestion in major cities around the world. One proposed solution to congestion or, better, to the underusage of private vehicles, is the so called “carsharing”, i.e., pools of vehicles to be rented for short periods of time (minutes, hours), usually at higher costs (per day) than standard car rental prices. In many urban areas, such as Milan, where the authors live, measures against congestion are combinedly applied, e.g., tolls to enter a particular area, carsharing (with access to the paying area included), dedicated lanes, ban for certain types (older ones) of vehicles. Carsharing vendors “publish” (not entirely/easily accessible) data about the state of their vehicle pool... Can this data be used to analyze these services’ effect, efficiency, usefulness, social cost, etc.? The authors scraped carsharing vendors’ websites for a year, made this huge amount of data uniform, fed it into a mongodb database and then “played” with queries and graphed results. An interesting finding is that even on the carsharing pool a “lung effect” (people moving-in in the morning, moving-out in the evening) is evident, i.e., the common notion that carsharing is not for commuters can be argued. Another interesting behaviour is the evening peak usage, i.e., probably, caused by people using carsharing instead of taxicabs to go out at night (leisure). Moreover, the data show that vehicle usage (the total number of “busy” vehicles at any time) never goes beyond 70%, i.e., there is always a 30% pool of “free” vehicles. Throughout the paper interesting statistical data and graphs will be shown and discussed.