基于K-Means聚类方法的北京市公共交通通勤乘客特征挖掘

Analysis of Passenger Characteristics of Steady-state Travel in Public Transportation in Beijing

北京市公共交通系统在满足居民出行需求的同时,产生海量的出行数据,为客流分析奠定数据基础,而地理信息技术、机器学习为数据处理与分析提供新的思路。基于2021年10月北京1周公共交通票务数据,通过多种清洗手段和补全算法对源数据进行处理,通过时间阈值和空间阈值提取出行链,从时间、空间、频次3个维度分析客流特性并构建聚类变量。采用K-Means算法将常乘客分为5类,分析各类别乘客特征规律,提取出稳态常乘客类别,为常乘客的分析提供新的思路,有助于界定公交的主要客流来源出行特征,为改善公交运营,提升公交服务水平提供支撑。

Beijing's public transportation system not only meets the travel needs of residents,but also generates massive travel data,which lays a data foundation for passenger flow analysis,while geographic information technology and machine learning provide new ideas for data processing and analysis.Based on the weekly public transport ticketing data in Beijing in October 2021,this paper processes the source data through a variety of cleaning methods and completion algorithms,extracts the travel chain by time threshold and spatial threshold,analyzes the passenger flow characteristics from three dimensions:time,space and frequency,and constructs clustering variables.The K-Means algorithm is used to divide frequent passengers into five categories,analyze the characteristics of each category of travelers,and extract the steady-state frequent flyer categories,which provides a new idea for the analysis of frequent passengers,helps to define the travel characteristics of the main source of bus passenger flow,and provides support for improving bus operation and improving bus service level.