Detecting arbitrarily shaped clusters in origin-destination flows using ant colony optimization |
| |
| Authors: | Ci Song Ting Ma Yunyan Du Hua Shu Sihui Guo |
| |
| Institution: | 1. State Key Laboratory of Resources and Environmental Information System, Institute of Geographical Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, China;2. College of Resources and Environment, University of Chinese Academy of Sciences, Beijing, China |
| |
| Abstract: | An origin-destination (OD) flow can be defined as the movement of objects between two locations. These movements must be determined for a range of purposes, and strong interactions can be visually represented via clustering of OD flows. Identification of such clusters may be useful in urban planning, traffic planning and logistics management research. However, few methods can identify arbitrarily shaped flow clusters. Here, we present a spatial scan statistical approach based on ant colony optimization (ACO) for detecting arbitrarily shaped clusters of OD flows (AntScan_flow). In this study, an OD flow cluster is defined as a regional pair with significant log likelihood ratio (LLR), and the ACO is employed to detect the clusters with maximum LLRs in the search space. Simulation experiments based on AntScan_flow and SaTScan_flow show that AntScan_flow yields better performance based on accuracy but requires a large computational demand. Finally, a case study of the morning commuting flows of Beijing residents was conducted. The AntScan_flow results show that the regions associated with moderate- and long-distance commuting OD flow clusters are highly consistent with subway lines and highways in the city. Additionally, the regions of short-distance commuting OD flow clusters are more likely to exhibit ‘residential-area to work-area’ patterns. |
| |
| Keywords: | Origin-destination flow spatial local statistics spatial scan statistics ant colony optimization |
|
|
