A stacking ensemble model based on Elastic Net and Random Forest for box office prediction

The rapid growth of the film industry has made accurate box office predictions crucial. A reliable prediction model can assist producers and insiders in optimizing resources by adjusting strategies based on forecast outcomes. This paper aims to compare the predictive capabilities of various models and proposes a stacking ensemble learning-based box office prediction model. We start by selecting relevant IMDB data factors such as movie duration, director details, cast lineup, and movie region for data cleaning and screening, thereby enhancing the input for box office predictions. Subsequently, we construct multiple forecasting models for comparison purposes and select the most superior ones as base models. We then create a secondary comparison stacking model. Furthermore, we utilize 10-fold cross-validation to fine-tune the model parameters, resulting in more precise evaluation results. Experimental findings demonstrate that XGBoost outperforms other single models, while different stacking ensemble models exhibit notable performance variations. The proposed stacking ensemble model, based on Elastic Net and Random Forest models, yields the best predictive outcome.