弹幕视频短期播放量预测及影响因素分析

Short-term Play Volume Forecast and Analysis of Factors Influencing Pop-up Videos

弹幕视频深受广大用户欢迎,通过预测弹幕视频短期播放量和影响因素的分析,便利视频平台判别视频质量高低并进行合理的推广安排,有利于提高平台弹幕视频服务和经济效益。抓取B站视频相关数据,将特征选择和不同算法结合分别构建随机森林模型、XGBoost模型和LSTM模型对弹幕视频播放量进行预测,比较分析不同特征组合进行预测对实验结果的影响。结果表明,随机森林模型预测精度要高于XGBoost模型和LSTM模型,且弹幕视频自身特征对播放量的影响最大,视频标文本特征对播放量的影响程度较小。

Pop-up videos are popular among users,and by predicting the short-term play volume of pop-up videos and analyzing the influencing factors,it facilitates the video platform to discern the video quality and make reasonable promotion arrangements,which is conducive to improving the platform pop-up video services and economic benefits. The data related to Beeping Beeping’s pop-up videos are captured,and the random forest model,XGBoost model and LSTM model are combined with different algorithms to predict the pop-up video playback,and the effects of different feature combinations on the experimental results are compared and analyzed. The results show that the prediction accuracy of random forest model is higher than that of XGBoost model and LSTM model,and the impact of pop-up video features on playback is the greatest,and the impact of text features such as video title and introduction on playback is smaller.