Research and Application of Carrier User Viscosity Evaluation Method Based on CNN Algorithm Model

Rui Ban, Hongding Wang, Yufei Hao, Salem Osman

Abstract: Carrier user viscosity directly reflects the data of user online behavior and user online quality. By studying and analyzing the status quo and existing problems of carriers user viscosity evaluation methods, this paper proposes a user viscosity prediction model design based on CNN algorithm, including technical scheme construction, sample data screening, algorithm model construction, etc. At present, the prediction model has been inputting real data on the live network, and the prediction results are compared and analyzed with real offline users on the live network. The accuracy and recall rate are good. Provide accurate user viscosity data for operators to facilitate business operation and market expansion.