Detecting Colocation Flow Patterns in the Geographical Interaction Data

The detection of colocation pattern is an important and widely used method to analyze the spatial associations of geographical objects and events. Existing studies primarily focus on discovering colocation patterns and association rules based on point data. A broad range of flow data types, such as population flow, logistics, and information flow, have emerged in recent years. However, colocation patterns and association rules based on flow data are difficult to detect because of their complex structure. This work proposes a colocation pattern detection and spatial association rule discovery approach that treats origin-destination (OD) flow as Boolean spatial features, while considering the spatial proximity of the origins and destinations of OD flows and its direction similarity. The effectiveness of this approach is verified by an artificial data set. Finally, this work analyzes the data of tourists who are traveling from different countries or regions to diverse cities in China. It also proves the application value of the proposed approach, which has general applicability to the mining of colocation patterns and association rules from any type of OD flow data.