KMAN Publication Inc. (KMANPUB) Journal of Adolescent and Youth Psychological Studies (JAYPS) 2981-2526 7 Serial Number 46 2026 06 10 LightGBM Classification of Academic Procrastination Risk Among Adolescents Based on Executive Function, Motivation, Smartphone Use, and Learning Strategy Variables LightGBM Classification of Academic Procrastination Risk Among Adolescents Based on Executive Function, Motivation, Smartphone Use, and Learning Strategy Variables 1 12 10.61838/kman.jayps.5445 EN Journal Article 2026 01 22 <table> <tbody> <tr> <td> <p><strong>Objective:</strong> The objective of this study was to develop and evaluate a LightGBM machine learning model for predicting academic procrastination risk among adolescents using cognitive, motivational, behavioral, and learning strategy variables.</p> <p><strong>Methods and Materials: </strong>A total of 1,248 adolescents aged 13–18 years from public secondary schools in Germany participated in this cross-sectional study. Participants completed standardized self-report instruments assessing executive function (Behavior Rating Inventory of Executive Function–Self-Report), academic motivation (Academic Motivation Scale), smartphone use patterns (Smartphone Addiction Scale–Short Version), and learning strategies (Motivated Strategies for Learning Questionnaire). Demographic information, including age, gender, grade, parental education, and academic achievement, was also collected. The dataset was preprocessed to handle missing data, outliers, and standardization, and academic procrastination scores were categorized into low, moderate, and high-risk groups. A LightGBM classification model was trained on 80% of the data using stratified sampling and five-fold cross-validation, with hyperparameter tuning performed to optimize model performance. Model evaluation was conducted on the remaining 20% of the data using accuracy, precision, recall, F1-score, AUC-ROC, and interpretability analysis via SHAP values.</p> <p><strong>Findings: </strong>Executive function difficulties, low academic motivation, problematic smartphone use, and ineffective learning strategies were significant predictors of academic procrastination. Metacognitive self-regulation, working memory difficulties, intrinsic motivation, smartphone dependence, and time management emerged as the most influential features. The LightGBM model achieved an overall accuracy of 88.7%, macro-averaged F1-score of 0.877, and an AUC-ROC of 0.931. Misclassifications primarily occurred between adjacent risk categories, indicating the model’s robustness in distinguishing different levels of procrastination risk. SHAP analysis confirmed the relative contributions of cognitive, motivational, behavioral, and learning strategy factors in shaping predicted risk levels.</p> <p><strong>Conclusion: </strong>The study demonstrates that an interpretable LightGBM model can accurately classify adolescents according to academic procrastination risk based on executive function, motivation, smartphone use, and learning strategy variables, highlighting the critical roles of cognitive self-regulation, motivation, and digital behavior in academic outcomes.</p> </td> </tr> </tbody> </table> https://journals.kmanpub.com/index.php/jayps/article/download/5445/10596