KMAN Publication Inc. (KMANPUB) Applied Family Therapy Journal (AFTJ) 3041-8798 7 Serial Number 33 2026 06 01 Explainable LightGBM Analysis of Adolescent Family Adjustment Based on Parenting Style, Family Communication, Emotional Security, and Self-Esteem Explainable LightGBM Analysis of Adolescent Family Adjustment Based on Parenting Style, Family Communication, Emotional Security, and Self-Esteem 1 13 10.61838/kman.aftj.5647 EN Journal Article 2025 10 22 <table> <tbody> <tr> <td> <p><strong>Objective:</strong> The present study aimed to predict adolescent family adjustment based on parenting style, family communication, emotional security, and self-esteem using an Explainable Light Gradient Boosting Machine (LightGBM) model and SHapley Additive exPlanations (SHAP) analysis among Colombian adolescents.</p> <p><strong>Methods and Materials:</strong> This cross-sectional predictive study was conducted among 1,248 adolescents aged 13 to 18 years recruited from public and private secondary schools in Colombia through multistage cluster sampling. Data were collected using standardized measures of family adjustment, parenting style, family communication, emotional security, and self-esteem. Following data preprocessing and quality assessment, the dataset was randomly divided into training (80%) and testing (20%) subsets. A LightGBM algorithm was developed to predict family adjustment scores, and model hyperparameters were optimized using five-fold cross-validation. Model performance was evaluated using the coefficient of determination (R²), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). To enhance model interpretability and identify the relative importance of predictors, SHAP analyses were performed to examine both global and local feature contributions.</p> <p><strong>Findings:</strong> The Explainable LightGBM model demonstrated excellent predictive performance, accounting for 89.2% of the variance in family adjustment in the training dataset and 84.6% in the testing dataset (R² = .892 and .846, respectively). Cross-validation analyses confirmed model stability (Mean R² = .839, SD = .021). SHAP-based feature importance analyses revealed that emotional security was the strongest predictor of adolescent family adjustment, accounting for 34.8% of total predictive importance, followed by family communication (29.3%), authoritative parenting style (22.7%), and self-esteem (13.2%). SHAP summary analyses further indicated that higher levels of emotional security, more effective family communication, stronger authoritative parenting practices, and greater self-esteem were consistently associated with higher predicted family adjustment scores. The explainable machine learning framework successfully identified both the magnitude and direction of predictor contributions, highlighting the predominance of family relational factors in explaining adjustment outcomes.</p> <p><strong>Conclusion:</strong> The findings demonstrate that adolescent family adjustment can be accurately predicted through explainable machine learning approaches and that emotional security, family communication, authoritative parenting style, and self-esteem are significant contributors to adjustment outcomes.</p> </td> </tr> </tbody> </table> https://journals.kmanpub.com/index.php/aftj/article/download/5647/10614