基于Stacking集成学习的电影票房预测研究  

Prediction Research on Movie Box Office Based on Stacking Ensemble Learning

作　　者：姚雅琪 徐秀丽[1] 

机构地区：[1]燕山大学理学院,河北 秦皇岛

出　　处：《统计学与应用》2021年第2期193-208,共16页Statistical and Application

摘　　要：电影票房作为电影行业最为主要的收入来源,研究票房影响因素并对其进行预测,有利于电影行业的发展和投资者做出正确投资决策。该文结合人工智能的前沿理论研究,提出了一种基于Stacking集成学习的电影票房融合模型的预测方法。选取2017年至2019年票房排名前100的电影票房及相关影响因素数据进行分析,并清洗量化规约各影响因素,通过XGBoost算法计算特征重要性筛选主要影响因素;利用Stacking模型融合多个机器学习算法,构建电影票房预测模型,通过网格交叉验证优化模型参数,对比评估得Stacking集成学习模型较单个机器学习预测模型具有更好的预测效果,在电影票房预测方面有较高的应用价值。Film box office is the most important sources of income in the film industry. Researching the box office influencing factors and predicting them will help the development of the film industry and investors to make correct investment decisions. This article combines the cutting-edge theoretical research of artificial intelligence and proposes a prediction method of movie box office fusion model based on Stacking ensemble learning. The paper selects the box office data of the top 100 films and the relevant influencing factors from 2017 to 2019, cleans and quantifies influencing factors and uses the XGBoost algorithm to calculate feature importance to screen the main influencing factors. Then the paper uses the Stacking model to integrate several machine learning algorithms to build a box office prediction model, optimizes model parameters through grid cross- validation. Through comparative evaluation, it can be seen that the stacking integrated learning model obtained has a better prediction effect than a single machine learning prediction model, the model has high application value in movie box office prediction.

关 键 词：电影票房 集成学习 XGBoost Stacking模型融合 

分 类 号：J94[艺术—电影电视艺术]