手机信令聚类的个体常驻区域识别

Character recognition of individual resident area based on mobile phone signaling data mining

针对手机信令数据存在定位稀疏、精度粗糙以及语义信息有限等问题,该文提出了利用手机信令数据以识别个体常驻区域及停驻信息的方法,即采用滑动时窗去除数据噪声,计算生成用户的时序聚簇;再通过凝聚式层次聚类对长时间序列内的时序聚簇进行空间簇联合,推测个体的常驻区域范围;最后提取合并聚簇的时间信息,计算个体在相关常驻区域停驻的时段特征。实验表明,该文方法有效识别了个体的常驻区域,降低了固定时段划分数据识别常驻地的局限性,细化地描述了个体的常驻区域及停驻时段特征。

Aiming at the problems of sparse positioning,rough precision and limited semantic information of mobile phone signaling data,this paper proposed a method to identify individual resident areas and parking information by using mobile phone signaling data.Firstly,the sliding time window was used to remove the data noise,and the user’s parking point clusters were calculated and generated.Then the agglomerative hierarchical clustering was used to combine the time-series clusters in the long-term series,and the range of the resident area of the individual was estimated.Finally,the time information of the merged cluster was extracted,and the time period characteristics of the individual staying in the relevant resident area were calculated.Experiments showed that the method in this paper can effectively identify the resident area of an individual,reduce the limitation of identifying the resident place by dividing the data in a fixed period of time,and describe the resident area and the characteristics of the resident period of the individual in detail.