SCALING PATTERNS OF NATURAL URBAN PLACES AS A RULE FOR ENHANCING THEIR URBAN FUNCTIONALITY USING TRAJECTORY DATA

Abstract. With the availability of massive trajectory data, it is highly valuable to reveal their activity information for many domains such as understanding the functionality of urban regions. This article utilizes the scaling patterns of human activities to enhance functional distribution of natural urban places. Specifically, we proposed a temporal city clustering algorithm to aggregate the stopping locations into natural urban places, which are reported to follow remarkable power law distributions of sizes and obey a universal law of economy of scale on human interactions with urban infrastructure. Besides, we proposed a novel Bayesian inference model with damping factor to estimate the most likely POI type associated with a stopping location. Our results suggest that hot natural urban places could be effectively identified from their scaling patterns and their functionality can be very well enhanced. For instance, natural urban places containing airport or railway station can be highly stressed by accumulating the massive types of human activities.