Exploring the Major Predictors Affecting Creativity of High School Students Using Explainable Artificial Intelligence: Application of Random Forest and SHAP

This study was designed to explore major predictors related to creativity of high school students to discuss the implication for creativity development. To do this, random forest was applied to 5th wave data(11th graders) of the Korean Children & Youth Panel Survey(KCYPS) 2018, and explainable artificial intelligence SHAP(Shapley addictive explanations) analysis was conducted to explore the importance of major predictors and the relationship between creativity and the predictors. The main result are as follows. First of all, not only individual-related factors but also family-, school-related and academic factors were derived as main predictors for creativity. Among the individual-related factors, ‘self esteem’ was derived as the most important factor. Additionally. ‘academic enthusiasm’, ‘academic engagement’ were found as important factors. Among family-related factors, ‘parental attitude’ was derived as important predictors., Among school-related and academic factors, ‘youth activities participation’, ‘club activities participation’, ‘self study time’, ‘academy and tutoring time’ were derived as important predictors. Based of these findings, implication for developing creativity and suggestions for future research are discussed.