电影首映日后票房预测模型研究  

Box Office Forecasting after Premiere

作　　者：罗晓芃 齐佳音[2] 田春华[3] 

机构地区：[1]北京邮电大学经济管理学院,北京100876 [2]上海对外经贸大学工商管理学院,上海201620 [3]清华大学软件学院,北京100084

出　　处：《统计与信息论坛》2016年第11期94-102,共9页Journal of Statistics and Information

基　　金：国家自然科学基金重点项目<面向不确定性的Web2.0用户创作内容管理研究>(71231002);国家自然科学基金重大研究计划培育项目<大数据驱动的三类典型工商市场主体风险识别决策研究>(91546121)

摘　　要：利用285部电影的截面数据进行单样本t检验,发现票房过亿电影的首映日票房都不少于200万元。抽取138部首映日票房超过200万元的电影的21天数据构建动态面板模型,采用两步系统GMM估计建立票房预测模型。研究表明:前一日票房对后一日票房具有显著正向影响,即在一段时间内前一日票房增加将提升后一日票房。票价较高和较低都会对票房产生积极影响,因此将票价纳入票房预测模型中。电影类型、上映日期、上映档期、出品国别、续集、网络评分、点映活动均被证明对票房具有正向影响。相较其他研究,包含上述指标的预测模型准确率得到了大幅提高。Based on the cross-sectional data of 285 films,we found that the movie whose premiere box office exceeds two million Yuan could often attains over one hundred million Yuan in total box office.Then we extracted 138 films whose premiere box office was over two million Yuan,and created a dynamic panel data sample with 21 release days＇ data,established a box office forecasting model by two-step system GMM estimation method.As a result,the previous day＇s box office has a significant and positive impact on the next day＇s box office,and ticket price will affect total box office when the price is relatively high and low.Genre,releasing weekday and season,produced country,sequel,rating score,and early small-scale releasing are proved to influence box office.Combined with these factors above,the box office forecasting model has achieved much higher accuracy than other research.

关 键 词：首映日 票房预测 动态面板 广义矩估计方法 

分 类 号：J943[艺术—电影电视艺术] O211.67[理学—概率论与数理统计]