摘要
针对马尔可夫链预测的局限性,本文提出了能够定量计算叠加4G流量订购量的马尔可夫链预测方法。以定西地区2月4G流量订购量进行状态分级,划分为滞销、偏滞销、一般、偏畅销和畅销,计算状态转移矩阵,将不同步长转移矩阵求得的预测值进行加权平均,并运用叠加马尔可夫链方法建立该地区4G流量订购量模型,分析拟合结果。结果表明,该模型的预测精度达到了87.99%,预测效果较好,为4G流量订购量的预测提供了一种方法。
Aiming at the limitation of Markov chain prediction,a Markov chain forecasting method is proposed to calculate and superpose the order amount of the4G roaming data quantitatively.In the Dingxi prefecture’s order amount of the4G roaming data in february for example,it is classified by status,which includes fi ve parts,the unsalable,the partial unsalable,the general,the popular and the bestselling.The model of the region’s4G roaming data ordering amount is established by using the additive Markov chain method,through weighted averages of the predicted value obtained by the different step transfer matrix.Last but not least,the results show that the prediction accuracy of the model is up to87.99%,and the prediction effect is good,which provides a method for the prediction of4G traffi c order quantity.
作者
任小强
夏耩
吴光华
管一霖
REN Xiao-qiang;XIA Jiang;WU Guang-hua;GUAN Yi-lin(China Mobile Group Gansu Co., Ltd. Lanzhou Branch, Lanzhou 730000, China;Chengdu University of Technology,Chengdu 610059, China)
出处
《电信工程技术与标准化》
2017年第8期82-85,共4页
Telecom Engineering Technics and Standardization
基金
2016年国家级大学生创新创业训练计划项目(201610616091)
2016四川省科技创新苗子工程(2016138)
关键词
叠加马尔科夫链
4G流量订购量
状态分级
superimposed Markov chain
business dealing of 4G data traffi c
status classifi cation