基于多元用户异常行为数据的用户分类模型研究与应用

Research and application of user classifi cation model based on multivariate user abnormal behavior data

对于电信运营商而言,挖掘海量用户的网络行为数据,提升数智化风险防控水平,具有重要意义。本文融合电信运营商的欠费行为数据和异常订购数据等风险数据源,通过特征提取、标准化处理与逻辑回归建模,构建更为精准且可迭代的用户分类模型,满足电信运营商在权益订购和发放等场景的业务风控需要,以实现对用户异常行为的有效预测和管理。实验结果表明,该方法为风控决策提供了有力的参考。

For telecom operators,it is of great significance to mine the network behavior data of massive users and improve the level of digital and intelligent risk prevention and control.This paper integrates risk data sources such as telecom operators'arrears behavior data and abnormal ordering data.Through feature extraction,standardization processing and logistic regression modeling,a more accurate and iterative user classifi cation model is constructed to meet the business risk control needs of telecom operators in scenarios such as equity ordering and equity distribution,so as to achieve effective prediction and management of users'abnormal behaviors.Experimental results show that this method provides a strong reference for risk control decisions.