Teodoro, Douglas Gülen, Lucas 2020 Carpooling is a mode of transportation that allows users to share places in their cars to reduce green-house gas emissions. Many services exist to encourage users to practice carpooling, while Mobilidée proposes a novel idea with a mobile application presented to other companies who want to promote soft mobility among their employees. Currently, Mobilidée already has an application that lets users enter manually whom they carpool with and for how long. The advantage for any user is that the more they reports their carpooling data in the application, the more they will have a chance to receive incentives from his company. Thus, to ensure security in this process as much as possible, Mobilidée needs to automate data collection and analysis of their users’ smartphone sensors in order to determine how much a given user has made carpooling in a given period. Indeed, today there is not a verification process in place that prohibits a user from claiming to carpool when, in reality, he does not. Hence, this work aims to investigate machine learning algorithms in order to classify people who made carpooling together or not automatically. This work will also explore some machine learning techniques to encode user data before their analysis to ensure their privacy. application/pdf https://sonar.ch/global/documents/315007 https://sonar.ch/documents/315007/files/Gulen-M_moire-Vdef.pdf eng info:eu-repo/semantics/openAccess License undefined mobilité carpooling application mobile mobilité douce info:eu-repo/classification/udc/02 Carpoolers classification while preserving users privacy http://purl.org/coar/resource_type/c_bdcc