Predicting Video Website DAU by Random Forest Algorithm

Abstract

With the rise of major video websites, the scale of video website users is expanding. Users are the core of all video websites, so it is of great significance to study video website DAU to evaluate and guide its operation. Based on the data of a video website DAU and platform layout from January 2017 to December 2018, this article analyzes the influencing factors of the video website DAU, selects 10 influencing factors from broadcasting arrangement, content level, content mode, and winter and summer vacations, constructs a Random Forest model, makes regression prediction on the daily average DAU of the video website and calculates the importance of 10 variables. The results show that the most important predictor of DAU is the self-produced drama, followed by the number of variety shows S-level, variety shows A-level and TV series. Video websites can use the method to predict DAU, and improve their operation from the above predictors.