您的位置:首页  > 论文页面

通信企业用户在网离网大数据分析与预测的实证研究

发表时间:2017-07-14  浏览量:2079  下载量:594
全部作者: 张维东,伍俊良
作者单位: 重庆大学数学与统计学院
摘 要: 近年来,大数据(big data)的研究与应用引起了学术界和企业的广泛关注,各级政府也从战略和政策层面给予了足够的重视和支持。本文从现代通信企业着手,利用大数据的理论基础与技术体系,对通信行业用户在网离网行为进行实证研究。尤其借助大数据处理方法对后付费移动用户的相关数据进行分析,构建移动用户在网离网的识别模型,预测用户离网风险情况。研究基于输入输出关联法(input output correlation,IOC)给出类二分法来提高建模效率。该方法不仅解决了径向基函数(radial basis function,RBF)人工神经网络建模优化过程中无法继续选择最优变量属性组合的问题,而且能够较快地选择最优变量,从而提高模型的精度。
关 键 词: 应用统计数学;大数据;理论基础;技术体系;通信;预测
Title: An empirical study on analysis and forecast of big data of on-line users and off-line users in communication enterprises
Author: ZHANG Weidong, WU Junliang
Organization: College of Mathematics and Statistics, Chongqing University
Abstract: In recent years, research and application of big data have been aroused wide attention of academia and enterprises. Governments have also given sufficient attentions and supports to it on the strategic and policy levels. This paper starts from the modern communication enterprises, uses the theoretical basis and technical system of big data to make an empirical study on the behaviors of on-line users and off-line users in the communication industry. In particular, the big data processing methods are also used to analyze the data of post-paid mobile users to build a mobile users’ on-line and off-line identification model, and forecasts users’ off-line risks. Based on the input output correlation (IOC), this paper gives the class dichotomy method to improve the efficiency of modeling. This method not only solves the problem that the optimal combination of variable attributes to be selected in the process of modeling optimization of radial basis function (RBF) artificial neural network, but also accelerates the speed of selecting the optimal variables, thus improves the accuracy of the model.
Key words: applied statistical mathematics; big data; theoretical basis; technological system; communication; forecast
发表期数: 2017年7月第13期
引用格式: 张维东,伍俊良. 通信企业用户在网离网大数据分析与预测的实证研究[J]. 中国科技论文在线精品论文,2017,10(13):1525-1533.
 
3 评论数 0
暂无评论
友情链接