Detecting Communities of Commuters: Graph Based Techniques Versus Generative Models

The main stage for a new generation of cooperative information systems are smart communities such as smart cities and smart nations. In the smart city context in which we position our work, urban planning, development and management authorities and stakeholders need to understand and take into account the mobility patterns of urban dwellers in order to manage the sociological, economic and environmental issues created by the continuing growth of cities and urban population. In this paper, we address the issue of the detection of communities of commuters which is one of the crucial aspects of smart community analysis. A community of commuters is a group of users of a public transportation network who share similar mobility patterns. Existing techniques for mobility patterns analysis, based on spatio-temporal data clustering, are generally based on geometric similarity metrics such as Euclidean distance, cosine similarity or variations of edit distance. They fail to capture the intuition of mobility patterns, based on recurring visitation sequences, which are more complex than simple trajectories with start and end points. In this work, we look at visitations as observations for generative models and we explain the mobility patterns in terms of mixtures of communities defined as latent topics which are seen as independent distributions over locations and time. We devise generative models that match and extend Latent Dirichlet Allocation (LDA) model to capture the mobility patterns. We show that our approach, using generative models, is more efficient and effective in detecting mobility patterns than traditional community detection techniques.