基于乘客上下车刷卡记录的公交客流时空分布研究

Fitting Analysis of Passenger Flow Distribution Based on Complete Passenger Transaction Records

基于乘客上下车刷卡交易记录,结合站点信息,有助于精准分析客流时空分布,为公交运营部门的调度决策提供参考。本文以北京市公交1路车为研究对象,针对公交客流在时间和站点上的多峰值分布特性,建立二维高斯混合模型描述客流分布,依据BIC准则确定聚类中心数K的值,选用EM算法对模型迭代求解。利用Python完成数据处理,结果表明,当K值取6时,对应模型的BIC值为300130.89,似然函数值为-149913.41,模型取得了较好的拟合效果,对分析客流时空分布和优化公交运营调度有较高的参考价值。

The passenger transaction records containing the information of getting on and off the bus,combined with the station information,can be used to analyze the spatial and temporal distribution of passenger flow accurately,and provide a reference for the scheduling decision of the bus operation department.Consequently,I took No.1 bus in Beijing as the research object and establish a two-dimensional Gaussian Mixture Model to describe the passenger flow distribution based on the multi-peak distribution characteristics of bus passenger flow in time and station.The number of cluster centers,K,was determined according to the Bayesian Information Criterion.I used Python to achieve the EM algorithm and solve the model iteratively.The results show that the model fits better when the K value is 6,the BIC value is 300130.89,and the likelihood function value is-149913.41.It can be a good reference for the bus operation.