Prediction of Mobile App Session Time

Abstract

With the rapid development of the mobile terminal applications, the research of mobile app advertising pricing model has attracted more attention than ever. Many researches draw a conclusion that the more advertisement exposure time, the more benefits they will get from the mobile applications. However, the current advertisement auction is priced and sold before actual viewing. In other words, the existing advertising pricing model doesn't involve the influence factor of the advertisement exposure time. Understanding the application usage patterns can help the advertisers to adjust the engagement of their ads. Therefore, this paper adopts a prediction model of the mobile application session time that combines the Back Propagation Neural Network and Genetic Algorithm. Finally, based on the experimental results and performance analysis, the error is in the acceptable scope.