A Method for Extracting Passenger Flow Time Series Feature of Urban Rail Transit

Urban rail transit passenger flow time series data has unique space-time feature. Using big data mining technology to master the feature is of great significance for ensuring station safety and improving service level. The main purpose of this paper is to propose a method for extracting passenger flow time series feature which collected by AFC, video surveillance, mobile phone signaling, WiFi probe technology, etc. Firstly, the agglomerate hierarchical clustering algorithm based on the average distance is used to classify the existing historical data. Moreover, an improved box plot analysis method is proposed to extract the characteristic curves of various types of data as feature representations. In addition, considering the efficiency of data update, a long-term and short-term combination update strategy is proposed. In the end, the method is practiced on the passenger flow data of Hanzhong Road Station of Shanghai Metro.