时间序列数据挖掘:公共自行车系统时空聚类  

Time Series Data Mining:Spatio-temporal Clustering of Public Bicycle System

作　　者：汪丽娜[1,2] 申江龙 WANG Li-Na;SHEN Jiang-Long(College of Sciences,Inner Mongolian University of Technology,Hohhot,Inner Mongolia 010051,China;Inner Mongolian Key Laboratory of Statistical Analysis Theory for Life Data and Neural Network Modeling,Hohhot,Inner Mongolia 010051,China)

机构地区：[1]内蒙古工业大学理学院,内蒙古呼和浩特010051 [2]内蒙古自治区生命数据统计分析理论与神经网络建模重点实验室,内蒙古呼和浩特010051

出　　处：《内蒙古工业大学学报（自然科学版）》2020年第4期249-255,共7页Journal of Inner Mongolia University of Technology：Natural Science Edition

基　　金：国家自然科学基金项目(11861049);内蒙古自治区自然科学基金项目(2018LH01012)。

摘　　要：共享经济浪潮下,自行车再次成为重要的城市交通工具.依据公共自行车系统运营产生的海量用户骑行数据,文章以自行车净流出数为聚类指标运用K-mean算法实现了自行车站点的时序特征聚类和集群的空间分布展示.构建了基于自行车净流出数的新聚类指标,该指标不依赖于站点的物理情况和自行车的物理状态;可以从微观上刻画站点的流量变化;可以体现集群的潮汐效应.通过对Citi Bike系统自行车站点的聚类分析,可以发现:住宅型站点和工作型站点具有相反的潮汐现象;两类运输型站点的流量近似动态平衡且具有微小相反的差异.通过聚类分析,能够发现站点的使用模式特征,为优化站点布局、调度车辆提供决策依据.Riding on the high tide of the sharing economy,bicycles have once again become an important urban transportation tool.Based on the massive user cycling data generated by the operation of the public bicycle system,this paper uses the net outflow of bicycles as the clustering index and uses the K-mean algorithm to realize the temporal feature clustering of bicycle stations and the spatial distribution of the clusters.A new clustering indicator based on the number of net outflows of bicycles is constructed.This indicator does not depend on the physical condition of the site and the physical state of the bicycles;it can microscopically describe the changes of the flow at the site;and it can reflect the cluster's tidal effect.Through the cluster analysis of the bicycle stations of the Citi Bike system,it can be found that the residential stations and working stations have opposite tidal phenomena,with the flows of the two types of transportation stations approximately being in dynamic balance but with slight reverse differences.Through cluster analysis,it is possible to discover the characteristics of the usage pattern of the stations,which will provide a basis for decision-making to optimize the layout of the stations and the dispatch of vehicles.

关 键 词：聚类分析 时间序列 数据挖掘 城市交通 公共自行车 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论] U121[自动化与计算机技术—计算机科学与技术]