User habits analysis in a bike sharing system

The concept of mobility is changing. Nowadays there are different systems to help people move in a sustainable way. The concept of sharing vehicles like cars or bicycles is more and more spreading. The bike sharing is growing up in the last years in many cities in different parts of the world. People prefer bikes because they are a healthy and a fast way to move especially across small distances. Moreover, people enjoy riding a bike. There are two types of bike sharing: the station-based bike sharing and the free-floating bike sharing. The first one is usually public, the user has to do the registration on the web site and then the service can be used. But it is based on station and this means that the user must take and park the bike in a station that must be found through the mobile app. The free-floating bike sharing is a relatively new system. The most important feature is that docking stations are not needed and the cost of installing and maintaining the racks are avoided. For this reason, the cost of the installation for the companies is less than the station-based system. Thanks to this new system searching a station to park the bike is not necessary, hence it is easier for the user to leave the bike. In the last few years not a lot of studies were made on the bike sharing system especially for what concern the free-floating bike sharing. The aim of this master thesis is to analyse the data coming from a free-floating bike sharing system to know the habits of the customers. To this purpose two types of data research are made: Real time data: to access this type of data some APIs are tested in order to have in real time the data of users’ movements in a certain city; Historical data: the website of Deutsche Bahn makes available a dataset of a bike sharing system owned by them named "Call A Bike". That dataset contains the historical data of the users in many German cities. For the first approach the Ofo system in the city of Milan considered. Data are retrieved from a server thanks to a POST request and at the same time a crawler is built up in order to process the data. But in this case some issues with the collection of data occur such as the coverage of the city, in fact the requests give back all the bikes present in a certain radius of a given point, hence the division of the city is not trivial and the token used for the request has a deadline, so a lack of data occurs. For the second approach a dataset of the Call A Bike system is found from the Deutsche Bahn website. From these data four main cities are selected (Munchen, Stuttgart, Frankfurt and Hamburg) and the most important features are analysed in order to understand how the usage of the system changes in the different years.