Estimating the General Trend of the Danmaku in League of Legends eSports Based on Multivariate Regression

This paper proposes an estimating method for the general trend of the number of danmakus in League of Legends eSports, where the danmaku refer to real-time text comments sent by the audiences. Firstly, we collected 72 top-viewed League of Legends eSports videos and all their danmakus, calculated the average trend of in-game score superiority and the number of danmakus over time for all videos, and assumed that the changes in the number of danmakus were caused by the change in score superiority. Then, we innovatively extracted two major factors that affect audiences' psychology from the fitted average superiority funtion: competitive and suspense, and establish a binary linear model for estimating the number of danmakus. Finally, we verified the proposed model, and the results showed that it can correctly estimate the general trend of the danmaku in two kinds of typical games. This model could help game researchers or designers understand how the eSports games make audiences and players feel fun and then improve the game mechanics.