Research on the influencing factors of customer rating based on PCA and RBF neural network

Abstract

In the past four decades, China's mobile communication technology has made rapid development and application, changing the way of life of human beings and making it possible for everything to be connected. As the network continues to be built, the network coverage is getting better and better. Therefore, it is particularly important to study the factors that customers influence on the operator's product services and thus further improve the quality of network services. First, cleaning operations such as coding and missing value processing are performed on the data, and specific data are filled based on information gain to fully utilize customer information. Secondly, based on principal component analysis, the features of each influencing factor are dimensioned down, and the cumulative influence contribution rate of each principal component is calculated. The influencing factors that have a cumulative influence contribution rate of more than 80% in the two data affecting the scores are selected, and mathematical models of customer scoring based on relevant influencing factors are established for customer voice service and Internet access service respectively, and other customer scores are predicted accordingly. The models stabilized after 8 and 6 iterations for voice service and Internet service, respectively, and their mean square error was kept at about 2.