Kyonggi University, Suwon City, South Korea.
PLoS One. 2024 Oct 3;19(10):e0309227. doi: 10.1371/journal.pone.0309227. eCollection 2024.
To improve the accuracy and efficiency of box office prediction, this study deeply discusses the application of the optimized eXtreme Gradient Boosting (XGBoost) model in this scenario and its advantages compared with other commonly used machine learning models. By comparing and analyzing five models, involving the Deep Neural Network, Light Gradient Boosting Machine, Random Forest, Gradient Boosting Decision Tree, and CatBoost, several key performance indicators such as accuracy, precision, recall, F1 score, generalization error, stability, robustness, and adaptability score are comprehensively investigated. The research results reveal that the optimization model proposed in this study is superior to the comparison model in most evaluation indicators, especially when the data volume reaches 2500, showing obvious advantages. For example, the accuracy is increased to 0.9, the F1 score is 0.9, the generalization error is reduced to 0.09, and the stability score is as high as 0.98. The robustness and adaptability scores are both 0.97, which proves its strong prediction ability and high stability and robustness on large-scale datasets. Therefore, this study provides scientific data support and a decision-making basis for the film industry in the formulation of marketing and distribution strategies. Moreover, film producers and distributors can reasonably estimate their market performance early in film shooting, optimize investment decisions, and reduce economic risks through accurate box office predictions.
为了提高票房预测的准确性和效率，本研究深入探讨了优化后的极端梯度提升（XGBoost）模型在这种情况下的应用，以及与其他常用机器学习模型相比的优势。通过比较和分析五个模型，包括深度神经网络、轻量级梯度提升机、随机森林、梯度提升决策树和 CatBoost，综合研究了准确性、精度、召回率、F1 分数、泛化误差、稳定性、鲁棒性和适应性得分等几个关键性能指标。研究结果表明，与比较模型相比，本研究提出的优化模型在大多数评估指标上都具有优势，特别是在数据量达到 2500 时，表现出明显的优势。例如，准确性提高到 0.9，F1 分数提高到 0.9，泛化误差降低到 0.09，稳定性得分高达 0.98。鲁棒性和适应性得分均为 0.97，这证明了它在大规模数据集上具有强大的预测能力和较高的稳定性和鲁棒性。因此，本研究为电影行业在制定营销策略和发行策略时提供了科学的数据支持和决策依据。此外，电影制作人和发行商可以通过准确的票房预测，在电影拍摄早期合理估计其市场表现，优化投资决策，降低经济风险。
