Research On Telecom User Satisfaction Prediction

In recent years,the 5G commercialization process has been accelerating.In the era of rapid development of 5G,users have put forward higher requirements for operators' services.However,after the avalanche growth of the 4G market,the demographic dividend disappeared and the competition among telecom operators became increasingly homogenized.In order to pre-empt the upcoming 5G market,the top priority for the three operators is to focus on improving user satisfaction and reducing user churn.In order to improve user satisfaction more effectively,it is especially important to predict user satisfaction in advance.If you can predict whether the user is satisfied in advance,you can fix it in a targeted manner.In turn,it can significantly improve user satisfaction.This article is based on the user satisfaction forecasting business that the author participated in during an internship in Zhejiang,and compiled and supplemented it to form this article.In this paper,telecommunications user satisfaction is divided into three business processes: network quality,promotion activities,and tariff packages.Based on logistic regression,random forest and support vector machine,three machine learning algorithms are used to predict the user satisfaction of these three business processes.Firstly,the theory of machine learning algorithm is expounded.Then,the data set is preprocessed to prepare for the subsequent establishment of the prediction model.Then,logistic regression,random forest and support vector machine algorithm are used to train and predict the user satisfaction.Based on the results of logistic regression,the three business processes are merged to obtain the prediction model of the user's overall satisfaction.Finally,the main research conclusions are summarized and summarized,and the future research is prospected.The conclusions of this paper are as follows: machine learning is applied to user satisfaction predictions well;network quality business process should focus on improving network quality related indicators;expanding preferential efforts is conducive to improving user satisfaction;tariff package business process should focus on user data traffic cost.