Predicting potential customers of 5G services via ADTree

Abstract

Using customer history data and data mining methods to discover potential customers of new business has become an important means of precise marketing for telecommunication operators. In this paper, we use ADTree for potential 5G customers prediction. Firstly, the training data is preprocessed, including data cleaning, discretization and data transformation. Secondly, in order to improve the modeling speed and classifier performance, the information gain method is used to select attributes. Thirdly, ADTree is used to model the training data, and the accuracy rate of the model and AUC value are evaluated by 10 fold cross validation. Finally, this model is used to classify the test data, and general guidance rules are given. Experiments show that the prediction accuracy and AUC value of ADTree method are superior to other classification methods.