Evaluation of taxi carpooling feasibility in different urban areas through the K-means matter–element analysis method

Abstract

A K-means matter–element model for the evaluation of taxi carpooling feasibility was proposed. The model was established on the basis of the influential factors and characteristics of urban taxi carpooling and data of taxi traffic. The first step taken to improve the shortcomings of the existing element model is to determine the index interval of the matter–element characteristic quantity using the K-means clustering algorithm. The entropy method is used to determine index weight. Taxi carpooling feasibility can be evaluated when passenger point density, taxi passenger rate, taxi distribution density, average carpool kilometres, average carpool return and average successful carpooling index information are described. Results showed that the algorithm can be used to effectively evaluate the effects feasibility of taxi carpooling.