基于时空聚类的职住分析研究

Research on Jobs and Residence Based on Spatio-temporal Clustering

为了得到常驻居民的职住特征,有效区分常驻与非常驻群体,基于手机信令数据设计了得到识别居民职住地的分析模型。通过大量的基站信令的预处理,对轨迹的停留点、发生时间、频次等属性进行基于K-Means的时空聚类,通过一系列判定方法得到不同性质的常驻地。为了验证该模型的可行性,以陕西省某运营商提供的基站数据为例,识别出常驻人口的职住地,将其结果与已知用户职住地对比得出:距离误差在客观范围内,且能够投入正常使用。

In order to obtain the jobs and residential characteristics of the permanent residents,as well as to effectively distinguish between resident and non-resident groups,an analytical model for identifying the jobs and residential location is designed based on the mobile phone’s base station signaling data. Through the preprocessing of a large number of base station signaling,spatio-temporal clustering based on K-Means is performed on attributes such as stay points,occurrence time,and frequency of the trajectory,and a series of decision methods are used to obtain resident locations with different natures. In order to verify the feasibility of the model,taking the base station data provided by an operator in Shaanxi Province as an example,the permanent resident location of the resident population is identified,of which the results are compared with the known user resident locations:the distance error is within the objective range and can be put into normal use.