Abstract: | In order to effectively reduce the economic expenses for taxi companies and users, the analysis of satellite trajectory data for electrical taxis is used to compare and select suitable clustering methods for charging pile location planning. Focusing on the location planning of electrical taxi charging stations in Shanghai, the paper designed the outlier detection method based on isolated forest and clustering algorithms to clean up the taxi satellite data within the relevant time period, followed by data visualization processing.The clustering effects of five algorithms, including Agglomerative hierarchical clustering, GMM Gaussian mixed clustering, K-means clustering, Mean-shift clustering and Spectrum clustering were evaluated and compared. And the K-means algorithm was selected as the reference algorithm for charging pile location planning. From the perspectives of urban zoning and business operations, a site selection strategy is determined, which provides a foundation for the design and planning of the quantity and capacity allocation of electric taxi charging piles in Shanghai for the future. |