Movie Box Office Prediction Based on Multi-Model Ensembles

Abstract

This paper is based on the box office data of films released in China in the past, which was collected from ENDATA on 30 November 2021, providing 5683 pieces of movie data, and enabling the selection of the top 2000 pieces of movie data to be used as the box office prediction dataset. In this paper, some types of Chinese micro-data are used, and a Baidu search of the index data of movie names 30 days before and after the release date, coronavirus disease 2019 (COVID-19) data in China, and other characteristics are introduced, and the stacking algorithm is optimized by adopting a two-layer model architecture. The first layer base learners adopt Extreme Gradient Boosting (XGBoost), the Light Gradient Boosting Machine (LightGBM), Categorical Boosting (CatBoost), the Gradient Boosting Decision Tree (GBDT), random forest (RF), and support vector regression (SVR), and the second layer meta-learner adopts a multiple linear regression model, to establish a box office prediction model with a prediction error, Mean Absolute Percentage Error (MAPE), of 14.49%. In addition, in order to study the impact of the COVID-19 epidemic on the movie box office, based on the data of 187 movies released from January 2020 to November 2021, and combined with a number of data features introduced earlier, this paper uses LightGBM to establish a model. By checking the importance of model features, it is found that the situation of the COVID-19 epidemic at the time of movie release had a certain related impact on the movie box office.

Towards Data-Driven Decision-Making in the Korean Film Industry: An XAI Model for Box Office Analysis Using Dimension Reduction, Clustering, and Classification

“…Ni et al [ 49 ] aimed to predict the box office revenue of films in China. The authors collected data from ENDATA including 5683 pieces of movie data and selected the top 2000 pieces for the prediction dataset.…”

Explainable Neural Network analysis on Movie Success Prediction

“…They have collected the past 30 years data regarding of Indian movie information, especially all regional wise movies and their proposed model delivered 96% accuracy. Another article by Yuan Ni and colleagues using multimodel ensembles was published in 2022 as well [5]. Their final model outperformed XGBoost, LightGBM, and other models after being trained and tested on the dataset of the top 2000 movies.…”

DNS Request Log Analysis of Universities in Shanghai: A CDN Service Provider’s Perspective

“…XGBoost is widely applicable and has been used by researchers for different prediction scenarios, such as image classification [36], intrusion detection [37], malicious account detection [38], and cross-site influential user identification [39]. LightGBM optimizes the temporal and spatial performance based on the traditional GBDT algorithm to speed up the training of GBDT models without compromising the accuracy and has been applied in movie box office prediction [40], credit scoring [41], and network traffic classification [42].…”