基于时间戳间距的用户在线时长聚类方法

Method of user online duration clustering based on timestamp interval

在网络用户行为分析中,以时序维度为基础,研究用户网络行为的变化趋势,提出并挖掘更多有价值的信息,可为管理或商业决策提供有力支持。为此,文中提出一种基于时间戳间距的用户在线时长聚类方法,以用户访问日志文件中时间戳之间的间距作为特征,首先将获取的日志数据进行预处理,获得用户的在线时长统计;然后采用K-Means聚类算法对用户进行聚类,并使用轮廓系数对K值进行评价分析,确定聚类K值范围,准确判定用户单次访问在线时长及类型。采用真实校园网用户访问日志数据对所提方法进行评价,实验结果表明,该方法的准确度达到0.9180,精确度达到0.7685,召回率达到0.8093。

In the analysis of online user behavior,based on the temporal dimension,the changing trend of user network behavior is studied,and more valuable information is proposed and explored,which can provide strong support for management or business decision-making.On this basis,a method of user online duration clustering based on timestamp interval is proposed,with the interval between timestamps in user access log file as a feature.The obtained log data is preprocessed to obtain the statistics on user online duration.The K-Means clustering algorithm is used to cluster user,and the silhouette coefficient is used to evaluate and analyze the K value within the range,determine the range of K value,and accurately determine the online duration and type of user’s single visit.The proposed method was evaluated by using real campus network user access log data.The experimental results show that the accuracy of this method can reach 0.9180,the accuracy can reach 0.7685,and the recall rate can reach 0.8093.