Authors: Shi Xian*, Guangzhou University, Zhixin Qi, Sun Yat-sen University
Topics: Urban Geography, Transportation Geography, Geographic Information Science and Systems
Keywords: Social segregation, mobility, activity, i-STP, decision tree algorithms
Session Type: Paper
Start / End Time: 5:00 PM / 6:40 PM
Room: Madison A, Marriott, Mezzanine Level
Presentation File: No File Uploaded
Conventional studies on social segregation are mainly based on a static perspective that emphasizes neighborhood and residence. This paper reports an exploratory social segregation study conducted in Hong Kong that aimed to go beyond residential segregation using mobility–activity data, individual-level spatiotemporal proximity index, and decision tree algorithms. Base on a framework that involves four dimensions (spatial, temporal, mobility, and activity), this study paid special attention to the interactions among people with different socio-economic status, particularly in terms of variables such as housing tenure status, age, and occupation; and the constraints brought by their home location. The i-STP index was adopted to examine the temporal variations of social segregation at individual level in a week. Decision tree algorithms were employed to identify the mobility–activity patterns that exhibited the highest separability between different groups of people from a large number of features extracted from mobility–activity data. The Results indicated that the social segregation level decreases in the daytime and increases at night in general. This trend was found to be consistent across a week, adding empirical evidence to the argument that mobility may alleviate residential segregation. We also identified meaningful mobility–activity patterns with significant differences between different groups of people. The results are indicative of some social norms, including dining with family or friends, eating around the home, time of arriving home, young people or students being together with family, and some people hanging out alone. The findings of this study may provide implications for research in other areas and for policymakers.