An Improved Crowd Aggregation Prediction Algorithm Based on ARMA

Abstract

∗ The gathering of abnormal crowds has brought huge hidden dan-gers to public safety. Accurate prediction of abnormal crowd gathering can effectively prevent and reduce the risk of abnormal gathering, and support reasonable security response decisions. The traditional ARMA algorithm can only make smooth predictions based on past historical data, and cannot predict sudden crowd gathering events. In order to alleviate this problem, this paper proposes an improved ARMA prediction algorithm. By adding the important factor of activity events to perform regression analysis, the parameters of the traditional ARMA prediction algorithm can be adjusted and optimized, so that it can more accurately predict the abnormal clustering trend of people related to mass events in a designated area. The experimental results show the superiority of our algorithm.