Machine Learning in Short Video APP User Activity Prediction

摘要

In order to improve the accuracy and reduce the cost of forecasting, this paper uses machine learning related technology to solve this problem in the user activity prediction model of short video industry. Continuous use of short video APP by active users is a sufficient and necessary condition for its success. The prediction of user activity has a direct guiding effect on the subsequent user loss warning. Based on the analysis of the impact on user activity, this paper extracts the characteristics according to registration log, startup log, shooting log and behavior log, and proposes a prediction algorithm based on model fusion for user activity. Based on the experimental data, the results show that the predicted AUC value reached 0.9514.