Mining the Coopetition Relationship of Urban Public Traffic Lines Based on Time Series Correlation

Abstract

Abstract Along with the evolution of passenger flows within cities, the coordination between public traffic lines should be sustainably optimized with respect to the spatial distribution of the flow, though the lines were planned well at the beginning of the construction. It is critical to determine the coopetition between bus lines to optimize a transit network continuously. A method of mining coopetition relationship (MCBTC, Mining Coopetition relationship between Bus lines based on a Time series Correlation) based on passenger flow is proposed in this study. First, noisy, inconsistent or missing data are eliminated to obtain a passenger flow time series, and the proposed merging algorithm is used to extract the line passenger flow time series (LPFTS, Line Passenger Flow Time Series) by merging the passenger flow of adjacent buses from the same line. Then, to calculate the positive and negative correlation sequence sets, a clustering algorithm is proposed. The two sequence sets represent the competition and cooperation relationships, respectively. The MCBTC method has been tested with a practical data set, and the results show that it is very promising.