Between Crowds, Critics, and Budgets: Explaining Blockbuster Movie Success with Machine Learning

This study investigates the key determinants of blockbuster movie success using a comprehensive dataset that integrates information from IMDb, OMDb, Box Office Mojo, YouTube, and Twitter (currently X). Focusing on films released between 1990 and 2025, the analysis evaluates the predictive power of traditional factors such as production budget, audience ratings, and critic scores, alongside measures of online engagement and sentiment. A range of models is employed, including Elastic Net regression, Random Forest, and XGBoost, to capture both linear and nonlinear relationships. The results show that machine learning methods, particularly XGBoost, substantially outperform linear models in predicting box office revenue, with the best configuration achieving R2=0.939 and MAE=0.135. Across all models, production budget and audience engagement emerge as the strongest predictors of financial success, with social media variables contributing additional explanatory power. These findings highlight the dominant role of production scale and audience engagement even prior to movie release in driving revenue, as opposed to critics' reviews, which had little impact on performance.