KMAN Publication Inc. (KMANPUB) Journal of Assessment and Research in Applied Counseling (JARAC) 8 Serial Number 30 2026 07 01 Stacked Generalization Model for Predicting Happiness: Positive Affect, Gratitude, Social Connectedness, and Psychological Flexibility Stacked Generalization Model for Predicting Happiness: Positive Affect, Gratitude, Social Connectedness, and Psychological Flexibility 1 10 EN Journal Article 2026 01 17 <table> <tbody> <tr> <td> <p><strong>Objective:</strong> The present study aimed to develop and evaluate a stacked generalization machine learning model to predict happiness based on positive affect, gratitude, social connectedness, and psychological flexibility.</p> <p><strong>Methods and Materials: </strong>This cross-sectional predictive study was conducted on a sample of 742 adults recruited from Canada using stratified online sampling to ensure demographic diversity. Participants completed validated self-report instruments including the Subjective Happiness Scale (SHS), the Positive Affect subscale of the Positive and Negative Affect Schedule (PANAS), the Gratitude Questionnaire-6 (GQ-6), the Social Connectedness Scale-Revised (SCS-R), and the Acceptance and Action Questionnaire-II (AAQ-II). Data preprocessing involved handling missing values via multiple imputation, outlier adjustment using robust scaling, and feature standardization. The dataset was split into training (80%) and testing (20%) sets, and model training utilized 10-fold cross-validation. A stacked generalization model was constructed using base learners including linear regression, support vector regression, random forest, and gradient boosting, with a Ridge regression meta-learner. Model performance was evaluated using mean squared error (MSE), root mean squared error (RMSE), and coefficient of determination (R²), and interpretability was enhanced באמצעות SHapley Additive exPlanations (SHAP) analysis.</p> <p><strong>Findings: </strong>The stacked generalization model demonstrated superior predictive performance compared to all base models, achieving the highest explained variance (R² = 0.74) and the lowest prediction error (RMSE = 0.52). Among the predictors, positive affect exhibited the strongest contribution to happiness, followed by social connectedness, gratitude, and psychological flexibility. SHAP analysis revealed significant nonlinear and interaction effects among predictors, indicating that the combined influence of these variables enhances predictive accuracy beyond additive models.</p> <p><strong>Conclusion: </strong>The findings highlight the effectiveness of stacked generalization in modeling complex psychological constructs and underscore the central role of positive affect, gratitude, social connectedness, and psychological flexibility in predicting happiness, supporting the integration of machine learning approaches in psychological research and intervention design.</p> </td> </tr> </tbody> </table> https://www.journals.kmanpub.com/index.php/jarac/article/download/207/9677