Large agglomerations are faced with rising and increasingly diverse mobility demands, just as the need for sustainable mobility is also a growing focus. The joint evolutions of “urban rhythms” and “territories of daily living” are tending towards individualisation of mobility practices. Policy-makers seem to be addressing these issues, at least in the large agglomerations, by developing tools and applications aimed at informing users of mobility options and supporting them in taking account of real-time traffic conditions. However, policy-makers do not today have the tools for precise analysis of mobility practices in terms of choice of mode or itinerary based on more qualitative and “subjective” elements such as comfort levels or weather conditions. The project proposes to analyse mobility practices and in particular the choice of transportation mode and itinerary to understand user preferences and, ultimately, communicate more pertinently around the various possible alternatives. The goal is to use data mining and learning techniques to promote better knowledge of traffic conditions and better information for users. The aim is not to publish raw or even simplified data, but to transform it in order to aid the decision-making process, both individual and collective. For this purpose, the project team combines a practitioner partner (Grand Lyon) with specialists in transport economics and territorial development (LET), traffic modelling (LET and LICIT), geometrics (LET and LIRIS), data mining and learning (LIRIS and LICIT) and optimisation (LIRIS and LICIT). Together they will seek to design new tools able to extract knowledge from mobility data to better understand the dynamics of traffic in the city, based on elements of cost (temporal and/or monetary) as well as comfort. An innovative aspect of the project is the combination of two levels, individual and collective. Combining these two levels will help form a better understanding of the dynamics of the complex system of mobility in the city.