Scholarlink

<p>Accurate and effective forecasting in the film industry can enable better preparedness for serious issues such as economic bottlenecks and epidemics. Also, it can help marketing managers achieve substantial cost savings by optimizing theatre capacity utilization and film distribution strategies. Thus, improving forecasting accuracy using the right predictors for box-office data is a challenging problem. This study proposes a big data driven framework for forecasting top 10 grossing box office revenue in the North America (US, Puerto Rico, and Canada) using Google Trends (GT) data as a predictor that represents online consumer search behavior about cinema. A proposed deep learning Vanilla LSTM, a Gated Recurrent Unit (GRU), a baseline Linear Regression, benchmarking Extreme Gradient Boosting (XGBoost), Support Vector Regression (SVR), and Artificial Neural Networks (ANNs) are conducted for the analysis to prove the applicability of this framework. The findings suggest that GT is a statistically significant driver of box-office revenue. Besides, to determine the best forecasting model, two performance metrics were compared. Specifically, the findings indicate that Vanilla LSTM deep learning model improves forecasting accuracy over existing models.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Forecasting box office revenue using AI algorithms and consumer search behavior

  • 期刊论文

  • 发表日期：

    2025年11月18日

查看全文
  • Bahri Baran Koçak

摘要

Accurate and effective forecasting in the film industry can enable better preparedness for serious issues such as economic bottlenecks and epidemics. Also, it can help marketing managers achieve substantial cost savings by optimizing theatre capacity utilization and film distribution strategies. Thus, improving forecasting accuracy using the right predictors for box-office data is a challenging problem. This study proposes a big data driven framework for forecasting top 10 grossing box office revenue in the North America (US, Puerto Rico, and Canada) using Google Trends (GT) data as a predictor that represents online consumer search behavior about cinema. A proposed deep learning Vanilla LSTM, a Gated Recurrent Unit (GRU), a baseline Linear Regression, benchmarking Extreme Gradient Boosting (XGBoost), Support Vector Regression (SVR), and Artificial Neural Networks (ANNs) are conducted for the analysis to prove the applicability of this framework. The findings suggest that GT is a statistically significant driver of box-office revenue. Besides, to determine the best forecasting model, two performance metrics were compared. Specifically, the findings indicate that Vanilla LSTM deep learning model improves forecasting accuracy over existing models.