Multimodal mobility patterns in transit systems

Abstract

Public transportation systems are becoming more pertinent to the requirements of any country and hence they should be reliable and efficient. In this study, we try to model the waiting times of passengers that transit from trains to buses across complex multimodal systems in an efficient manner. Analysis will show how the characteristics of different MRT stations may ultimately affect the waiting times of commuters that transit from trains to buses. We propose spatial filtering to be used as our tool of analysis on the smart card data available. We apply wavelets on graphs to extract global and local variations in waiting times across the transit system of Singapore. The issues of model selection and interpretability of these methods will also be discussed. A key advantage of spatial filtering is that we do not need data from multiple time instances to infer the mobility patterns. Hence, the technique can be easily used for detection of anomalous conditions at MRT/bus terminals in real-time.