Data collected by Automated Fare Collection (AFC) systems are a valuable resource for studying the travel habits of large city inhabitants. In this paper, we present an approach to mining the temporal behavior of the passengers in a public transportation system in order to extract relevant and easily interpretable clusters. Such classification can be useful for several applications. It may help transport operators better know the demand of their customers and propose targeted incentives, services, and tools accordingly. From a city perspective, this may also help redesign and improve existing transportation policies. To achieve this objective, an additional step of analysis is required and the clustering results need to be contextualized. The spatial location of the different types of passengers is then of great interest. We propose a first step in this direction through a rough estimation of the regular passengers' 'residence' location and the analysis of socioeconomic information available at a fine-grained spatial level. The approach is applied on a real dataset from the metropolitan area of Rennes (France) with four weeks of smart card data containing trips made by both bus and subway.