Spatial data mining and O-D hotspots discovery in cities based on an O-D hotspots clustering model using vehicles' GPS data: a case study in the morning rush hours in Beijing, China

With the rapid development of cities in recent years, the size of the cities is becoming bigger and bigger and the structure of the cities is becoming more and more complex. The first step to study the urban resilience is hotspots mining and POI analysis. This paper established an O-D hotspots clustering model based on Iterative Self Organizing Data Analysis Techniques Algorithm (hereinafter referred to as ISODATA) to mine the Origin-Destination (hereinafter referred to as O-D) hotspots in the rush hours in cities and study the distribution characteristics of Point of Interests (hereinafter referred to as POIs) in the hotspots area. It is found that the pick-up hotspots tend to be gathered in the residential zones and the drop-off hotspots tend to be gathered in the working zones. Besides, the distribution characteristics of POIs in both pick-up and drop-off hotspots areas and huge railway stations (special drop-off hotspots) areas are quite special. This study provides an in-depth understanding of the structure of the cities and provides an effective guidance in urban zones planning. This study also provides fundamental knowledge for urban resilience design.