Hao Cheng, Xiaoling Lv, Chao Fan, Yu Zhao

【Abstract】

To avoid the limitations of sensitive information on users’ behavior features and keep pace with mobile intelligence, this paper comes up with a new thought marked by APP application program and employs decision tree and other seven classification models to depict users’ behavior features and forecast choice preference. Study result shows that decision tree builds a relationship between mobile attributes and APP application and finds the law of user behavior features. Through comparative analysis on the classification accuracy of choice preference to the smart-phone APP of different types, the paper selects a model with best performance in forecast--Random Forest and Bagging, which provides a methodology foundation from smart-phone development stage to its marketing links.

【Keywords】

smartphone, APP application, users’ behavior features, forecast choice preference