关键词:
用户体验
Stacking
Adaboost
满意度预测
摘要:
随着移动通信技术的飞速发展,电信用户群体数量不断的攀升,也越要求运营商重视用户使用体验,不断提升网络使用满意度。本文基于2022年北京移动提供的客户语音和上网业务数据,首先使用灰色关联分析筛选出满意度重要影响因素,然后采用Stacking多模型融合策略,结合使用随机森林、逻辑回归、K近邻、Adaboost、XGBoost共5种算法,对客户满意度打分进行预测研究,融合后模型在语音业务数据中预测准确率为0.613,在上网业务数据中预测准确率为0.606,为电信用户满意度评分预测和分析研究提供了一定的理论参考。With the rapid development of mobile communication technology, the number of telecommunication user groups continues to increase, and the more operators are required to pay attention to user experience and continuously improve the satisfaction of network usage. This paper is based on the customer voice and Internet service data provided by Beijing Mobile in 2022, first we use grey correlation analysis to filter out the important factors affecting satisfaction, and then we adopt the stacking multi-model fusion strategy, combined with the use of Random Forest, Logistic Regression, K Nearest Neighbours, Adaboost and XGBoost in a total of five algorithms, to carry out prediction research on customer satisfaction scoring. The prediction accuracy of the model is 0.613 in voice service data and 0.606 in Internet service data, which provides certain theoretical reference for the prediction and analysis research of telecom customer satisfaction scoring.