Gaussian process-based predictive modeling for bus ridership.

ABSTRACTThe dynamics of a city are characterized, among others, by the traveling patterns of its dwellers. Accurate knowledge of human mobility patterns would have applications, e.g., in urban design, in the optimization of public transportation operating costs, and in the improvement of public transportation services. The present paper combines a large scale bus transportation dataset with publicly available data sources to predict bus usage. We propose a Gaussian process-based approach for modeling and predicting bus ridership. To validate our approach we perform experiments on data collected from Lisbon, Portugal. The results demonstrate significant improvements in prediction accuracy compared to a probabilistic baseline predictor.