A Novel Grid Based K-Means Cluster Method for Traffic Zone Division

Abstract

Traffic zone division plays an important role in analyzing traffic flow and the trend of city traffic. A traditional method based on sampling investigation has the shortcomings of high cost, long period and low sampling precision. With the development of traffic control and management methods, some cluster methods for location points are proposed to be used in the division of traffic plot. However, simple clustering analysis often need to detect the boundary of traffic zones, and the boundary of the division result is not clear, furthermore, abnormal data has great influence on results. In order to make the traffic zone division results clear and accurate and reduce the cost of the division, this paper proposes a novel grid based K-Means cluster method for traffic zone division by using Taxi GPS data. The experiment used GPS data of Nanjing city taxies and automatically divided traffic zones in Nanjing City area, which verified the validity of this partition method. The experimental results show that this classification method is effective and gives a good reference value for the analysis of city traffic flow and trends.