Authors: Luning Li*, Beijing Normal University, Qiang Li, Beijing Normal University
Topics: Behavioral Geography, Transportation Geography
Keywords: open public place, crowd risk, Wi-Fi probe, space-time activity
Session Type: Paper
Start / End Time: 8:00 AM / 9:40 AM
Room: Council Room, Omni, West
Presentation File: No File Uploaded
Open public place with specific function (e.g park) is densely populated with a diverse mix of people and thus characterized by high crowd risk. Population segmentation classifies population clusters by exploring activity patterns of distinct people, which is helpful to make the appropriate strategy for congestion alleviation. Shichahai scenic area in Beijing city is a typical open public place with multi-functions including tourist attraction, business and residence. Wi-Fi probe, one of the emerging position sensing devices, can conveniently collects the spatial-temporal information of individual activity through detecting the media access control (MAC) address of mobile devices. However, the spatial-temporal data collected by Wi-Fi probe is huge and difficult to distinguish population. A new spatial-temporal analysis approach is developed in the study. Firstly a m×n space-time matrix is constructed as m time intervals for observations of crowd flow, and n sites of Wi-Fi probe installation with different land-use functions. The POI data are applied to identify the land-use function of each site through semantic analysis method. Activity records of each individual are input into the space-time matrix, which contains the activity information in dimensions of time and space. Then, K-Means algorithm is conducted to segment the population into residential, staff and visitors, respectively. The results of the proposed approach agree well with field survey conducted on May 1-7, 2017, with overall accuracy of 87%-95%, suggesting that the proposed method is effective for population segmentation and valuable for making public safety strategy in open public places.