Time-Space Domain of Group Behavior Research Based on Major Events Influence

摘要

When a major event creates a behavior cluster, this cluster's spatiio-temporal data is the group behavior time-space domain. The time data of each individual on the space is a time series data. So this article we first to analyze the time series, approximately express the time series with the piecewise linear representation (PLR) method, and then use activity measure method for abnormal detection to get the abnormal time domain. According to the time of the major events occurred, we match individual abnormal time domain, and get the common abnormal groups. Finally, for these abnormal groups, we can get the space domain of the event with k-Medoids clustering algorithm analysis.