Data-efficient creativity evaluation in museum cultural creative products: a machine learning framework for data-driven decision-making in product development

This study addresses a critical gap in the evaluation of Museum Cultural and Creative Products (MCCPs), where existing models, such as the Museum Product Creativity Measurement (MPCM) model, though effective, are often too complex and impractical for real-world application, especially when supporting data-driven decision-making in product development. The research investigates whether the MPCM model can be simplified without compromising its predictive accuracy and explores the most suitable machine learning algorithms for creativity prediction. The study consists of two phases and utilizes a comprehensive dataset containing 5,423 participants and 17,853 data points from four distinct sources. In the pilot phase, data were collected through online and offline surveys, resulting in the development of three models: the Expert Suggested Model, the Hybrid Opinion Model, and a Machine Learning Model. The in-depth phase involved evaluating five machine learning models—Random Forest (RF), Gradient Boosting Decision Trees (GBDT), Light Gradient Boosting Machine (LightGBM), Support Vector Regression (SVR), and eXtreme Gradient Boosting (XGBoost)—using statistical analysis, model validation, and cross-validation techniques. The RF model underwent four rounds of testing, consistently demonstrating superior performance compared to the MPCM model, especially in predicting creativity with smaller sample sizes (200–300), with average RMSE and MAE values of 0.127 and 0.111, respectively. It indicates a notable difference between consumer-rated and RF-predicted creativity. This research contributes to the theoretical advancement and practical streamlining of creativity evaluation frameworks, enhancing their applicability to MCCPs across diverse cultural contexts. Furthermore, it offers methodological insights into how data-driven approaches inform and enhance decision-making processes in product development.