Analysis of car sharing data and service availability prediction using data mining techniques

The topic of this thesis is the analysis of the car sharing service, in particular of the analysis of the availability of this kind of service in terms of space (i.e, different areas of a city) and time (i.e., different moment of time, days of the week, slots of time and so on). In order to do so I exploited several regression algorithms, that gave as their outputs the prediction of the number of cars available in a given area in a given moment of time. The analysis of the prediction error of these algorithms allowed to determine which are the situations, from both space and time point of view, in which the service is most used. Such analysis allowed also to determine which are the techniques that give best performances when predicting this kind of data.