摘要
随着运营商家宽用户规模不断扩大,面对家宽网络的复杂性、用户侧组网的不透明性和家宽用户数据集高维不平衡等特点,传统的统计学方法已无法很好地满足家宽用户网络使用感知评估需求。本文提出一种基于SMOTE-PCA-RF模型的家宽潜在投诉用户识别模型,利用SMOTE算法与PCA算法进行数据预处理,用RF作为分类器。实验结果显示模型准确率为77%,可有效解决家宽用户满意度提升工作中遇到的“时效低、缺抓手”等难题。
As the scale of operators'home broadband users continues to expand,traditional statistical methods can no longer meet the complexity of home broadband networks,the opacity of user-side networking,and the high-dimensional imbalance of home broadband user data sets.Household broadband user network usage perception assessment needs,this paper proposes a household broadband potential complaint user identifi cation model based on the SMOTE-PCA-RF model,using SMOTE-PCA as the data preprocessing combination algorithm,and using RF as the classifi er.The experimental results show that the accuracy of the model is 77%.At the same time,the application results of the model show that the method in this paper can effectively solve the problems of"low timeliness and lack of hands-on"encountered in the improvement of home width user satisfaction.
作者
陈志安
杨慰民
CHEN Zhi-an;YANG Wei-min(China Mobile Group Fujian Co.,Ltd.,Fuzhou 350108,China)
出处
《电信工程技术与标准化》
2023年第7期16-21,共6页
Telecom Engineering Technics and Standardization
基金
中华全国总工会职工创新项目资助。
