基于随机森林回归模型的国产电影票房预测  

Research on Box Office Prediction of Chinese Films Based on Random Forest Regression Model

作　　者：李旺泽 郑列[1] LI Wangze;ZHENG Lie(School of Sciences,Hubei Univ.of Tech.,Wuhan 430068,China)

机构地区：[1]湖北工业大学理学院,湖北武汉430068

出　　处：《湖北工业大学学报》2020年第1期114-117,共4页Journal of Hubei University of Technology

基　　金：教育部人文社会科学研究规划基金(17YJA790098)。

摘　　要：以2011-2018年期间票房过亿的293部国产电影为样本,利用网络爬虫技术,从电影票房网、时光网、艺恩网、猫眼网以及微博选取电影观众评分人数、首日票房、首周票房、观众想看人数、明星的影响力、导演的影响力、编剧的影响力、制式、电影评分等因素为变量,选择其中的30部电影为测试集,263部电影为训练集,基于随机森林回归的方法,建立电影票房预测模型。研究结果表明,电影票房平均预测精度在85%左右。In this paper,293 Chinese films with a box office of more than 100 million yuan from 2011 to November 2018 were taken as samples.Indicators such as the number of movie audience rating,first-day box office,first-week box office,the number of audiences who want to see the movie,celebrity influence,director influence,scriptwriter influence,format and film score were selected from the five websites of ticket houses including mtime,art.com,maoyan.com and weibo by using crawler technology.Among them,263 were selected as training sets and 30 were test sets.The results showed that the average prediction accuracy was about 85%.Therefore,it has certain guiding significance to the risk control of film investment.

关 键 词：电影票房 随机森林回归 电影票房预测 

分 类 号：J943[艺术—电影电视艺术]