Analysis of Spatio-Temporal Feature Extraction and Hot Spot Detection Based on Didi Data

As a complementary taxi service, Didi is characterized by large traffic flows and mainly concentrated in urban centers. To study the spatio-temporal variation rules of residents’ activities and identify hot spots in travel, this paper takes Haikou City as an example to study the Didi orders from May 22 to 28, 2017. The DBSCAN algorithm and Kernel density algorithm are applied to identify and extract residents’ travel hot spots, and ArcGIS is used for visual expression. It is found that there are obvious morning, noon, and evening peaks on weekdays, meanwhile only noon and evening peaks on weekends. Hot spots weekends are more dispersed and diverse than on weekdays. Some areas are continuous hotspots, which do not change with time. The study reveals the travel rules of Didi passengers in Haikou, providing reference for reducing the empty driving rate and better vehicle scheduling.