电信客户欠费模型评估  

Telecom customer arrearages model evaluation

作　　者：包志强[1] 崔妍[1] 

机构地区：[1]西安邮电大学通信与信息工程学院,陕西西安710121

出　　处：《西安邮电大学学报》2015年第4期97-101,共5页Journal of Xi’an University of Posts and Telecommunications

基　　金：国家自然科学基金资助项目(61271276);陕西省自然科学基金资助项目(2012JQ8011)

摘　　要：针对客户恶意欠费对电信运营商造成的经济损失问题,提出一种基于BP人工神经网络的客户欠费模型,为新客户的欠费情况做出预测。以某电信运营商客户服务部数据为研究对象,CRISP-DM(跨行业数据挖掘过程标准)方法论为建模流程,采集2 500个样本进行网络训练测试,在允许误差为0.001时,预测率达到92.33%。实验结果表明,该模型能够较准确的预测客户欠费情况,提高了预测的有效性和实用性。A customers＇ arrearages model based on BP artificial neural network ts proposea mtne paper to target at the problem of customers＇ malicious arrearages resulting in economic losses for telecom operators. It provides prediction of the arrearage situation for the new customers. By using the data from a telecom operator customer service department and by CRISP-DM （Cross-Industry Standard Process for Data Mining） methodology for modeling process, a BP network model is set up. In this model, 2 500 samples are collected for network training and test, and with the allowing error 0. 001, the model＇s prediction rate can reach 92.33%. Experimental results show that the model can accurately predict the new customers＇ arrearages situation and improve the validity and practicability of the prediction.

关 键 词：BP人工神经网络 客户欠费预测 大数据 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]