In current wireless cellular system,a reasonable allocation of the cyber source is getting more and more important, and the foundation to reasonable resource allocation is a rational cognition for possible future needs, this cognition comes from our understanding of historical information and also comes from our grasp of user behavior for human.This paper mainly studies the prediction problem in traffic of wireless base station, in this study, we perform a statistical analysis on the raw data,mining useful information from the social network perspective hidden behind the traffic data, and then construct a traffic model to describe the problem. Based on the established model, we build two models of traffic prediction,one model is based on linear model with Lasso estimation, and the other one uses the Random Forest model to select useful items to improve the prediction precision of traditional autoregressive moving average mode. This work is based on an empirical data collected by a China Mobile company, and shows the effectiveness of the proposed scheme.Through the discussion and summary, we have formed a more stable and interpretable network traffic data analysis model. |