A Collective Human Mobility Analysis Method Based on Data Usage Detail Records.
Hao Jiang,Qian Li,Xian Zhou,Yanqiu Chen,Shuwen Yi,Hai Wang,Zheng Lu
DOI: https://doi.org/10.1080/13658816.2017.1370715
2017-01-01
International Journal of Geographical Information Science
Abstract:Human mobility patterns have been widely investigated due to their application in a wide variety of fields, for example urban planning and epidemiology. Many studies have introduced spatial networks into human mobility analyses at the collective level. However, these studies merely analyzed spatial network structure, and the underlying collective mobility patterns were not further discussed. In this paper, we propose a collective mobility discovery method based on community differences (CMDCD). We constructed spatial networks where nodes represent geographical entities and edge weights denote collective mobility intensity between geographical entities. The differences between communities detected from the networks constructed in different periods were then identified. Since collective spatial movement has a large influence on network structure, we can discover groups with different mobility patterns based on community differences. By applying the method to data usage detail records collected from the cellular networks in a city of China, we analyzed different collective mobility patterns between the Spring Festival vacation and workdays. The experimental results show that our method can solve these two problems of identifying community differences and discovering users with different mobility patterns simultaneously. Moreover, the CMDCD method is an integrated approach to discover groups whose mobility patterns have changed in different periods at the large spatial scale and the small spatial scale. The discovered collective mobility patterns can be used to guide urban planning, traffic forecasting, urban resource allocation, providing new insights into human mobility patterns and spatial interaction analyses.