手机用户上网时段的混合Markov预测方法

发布时间：2018-03-30 17:21

本文选题：手机上网数据　切入点：手机上网行为　出处：《地球信息科学学报》2017年08期

【摘要】：手机用户上网时段研究与预测对手机用户行为与模式分析、网络服务内容设计、网络黏性与心理、移动互联商业智能等具有重要意义。本文结合Markov模型和关联规则模型,提出一种手机用户上网时段的混合Markov预测方法——Lift-Markov(LM)方法,并采用中国某城市4G手机用户流量上网产生的流量收费数据进行实验验证与分析。研究发现:该实验区域37.66%的手机用户个体存在明显的以天为周期的周期性特性;本文所提出的LM方法在10、20、30、40、50、60 min间隔时的平均预测准确率都优于Markov模型和Mostvalue模型,其中在60 min间隔时能达到79.75%的平均准确率,优于Markov模型(74.64%)和Mostvalue模型(64.44%);LM方法的预测准确率分布相比于其他2种模型都要窄,而且密度分布峰值最高、标准差最小,说明本文方法对人群的上网时段预测准确率较为集中与稳定,具有较好的预测性能。
[Abstract]:The research and prediction of mobile phone users' Internet access period is of great significance for mobile phone users' behavior and pattern analysis, network service content design, network viscosity and psychology, mobile interconnection business intelligence and so on. This paper combines Markov model and association rule model, etc. In this paper, a hybrid Markov prediction method for mobile phone users is proposed, which is called Lift-Markov-LM. The experimental results show that 37.66% of the mobile phone users in this experimental area have a periodic characteristic of the period of the day. The average prediction accuracy of the LM method proposed in this paper is better than that of the Markov model and the Mostvalue model at the interval of 10 / 20 / 30 / 40 / 50 / 60 min, and the average accuracy is 79.75% at the 60 min interval, respectively. The prediction accuracy distribution of the LM method is narrower than that of the other two models, and the density distribution is the highest and the standard deviation is the smallest. The results show that the prediction accuracy of this method is concentrated and stable, and it has good prediction performance.
【作者单位】：武汉大学测绘遥感信息工程国家重点实验室;地球空间信息技术协同创新中心;中国移动通信集团湖北有限公司业务支撑中心;
【基金】：国家自然科学基金项目(41231171、41371420) 湖北省青年英才开发计划武汉大学自主科研项目拔尖创新人才类资助项目(2042015KF0167)
【分类号】：O211.62;TN929.53

【相似文献】