Abstract: It is becoming more and more important to make analysis on activity regularity of population in urban commercial districts, which would be beneficial to commercial infrastructure distribution and construction. The paper presents an approach to mining hot time periods in urban commercial districts using POI information and mobile phone data. First, a data field-based self-adaptive DBScan algorithm (DFB-saDBScan) is proposed to make clusters for commercial districts. Second, boundary determination algorithm is presented based on convex hull. Then, a K-means algorithm with constraints (cK-means) is proposed to make extraction of hot time periods from flow of people. On the other hand, we also propose an approach to regional POI proportion analysis via urban POI data. Taking advantage of structure measurement indexes of landscape ecosystem, we analyze relationship between population mobility and facility distribution in commercial districts. Taking a downtown area of Wuhan city, China as case study, our approach is verified to be effectiveness and accuracy.
Keywords: Point of Interest (POI), mobile phone data, self-adaptive DBS can algorithm (DFB-saDBS can), K-means algorithm with constraints (cK-means), hot time period analysis, facility proportion analysis
