Classification model for public-transport trip destinations based on density-peak clustering

In this study, to classify the purpose of public-transport trips in specific zones, we developed a model that uses smart card data to automatically classify the purpose of public-transport trips, based on a density-peak cluster algorithm. This model extracts features corresponding to different trip purpose and clusters groups having similar features. Based on the results, we can determine the average feature of each trip feature cluster and, based on the trip features, obtain statistical analyses for the trip purpose of every passenger and group. Using data obtained from the smart public-transportation cards of Xidan regional travelers in Beijing, we compared and verified actual data with questionnaire results. The results show that, compared to the questionnaire approach, the model saves considerable manpower and material resources, obtains good results, and thus has practical value.