A Modeling Framework for Individual-Based Urban Mobility Based on Data Fusion

Abstract

Modeling individual-based urban mobility plays an important role in traffic management, urban planning, public health, public safety and many other fields. Compared with census and travel survey data, which are costly in collection and slow in update, the emergence of massively and automatically generated individual trajectory data, such as mobile phone tracking data and transit smart card data, offers new datasets to develop individual mobility models. However, these new types of human trajectory data suffer some inherent limitations for many research and application domains. First, these large-scale trajectory data often have certain types of sampling biases in the representation of entire population. For example, mobile phone data do not likely cover children and have little coverage in elder people. However, this portion of population is important in some studies, such as household-based travel demand modeling and epidemic modeling. Second, these large-scale individual trajectory data do not often come with individual sociodemographic attributes due to privacy or technical issues. Sociodemographic attributes however are critical in many studies such as household-based travel demand modeling, sociology studies, and epidemic modeling. Therefore, this study proposes a generalizable modeling framework for individual-based urban mobility with entire population and sociodemographic details, through integrating different types of data sources. To demonstrate the proposed modeling framework, we select several typical data sources, design a set of data fusion algorithms, and simulate the daily activities and trips of the entire population in Shenzhen, China. The simulation results show that the proposed data fusion approach can effectively help with sampling bias issues and reasonably fill up sociodemographic details for the large-scale trajectory data. The proposed individual-based urban mobility model can be useful in many studies that require inputting entire population with sociodemographic attributes. This study also gives an example of addressing an important topic in the “big data” era, that is to integrate the so called “big data” and traditional data in urban studies.