KMAN Publication Inc. (KMANPUB) Journal of Adolescent and Youth Psychological Studies (JAYPS) 2981-2526 7 Serial Number 45 2026 05 01 Predicting Romantic Relationship Instability in Emerging Adults Through LightGBM Models of Attachment Anxiety, Jealousy Cognitions, Social Media Surveillance, Emotional Dependency, and Impulsivity Predicting Romantic Relationship Instability in Emerging Adults Through LightGBM Models of Attachment Anxiety, Jealousy Cognitions, Social Media Surveillance, Emotional Dependency, and Impulsivity 1 14 10.61838/kman.jayps.5444 EN Journal Article 2025 09 17 <table> <tbody> <tr> <td> <p><strong>Objective:</strong> The present study aimed to predict romantic relationship instability in emerging adults through Light Gradient Boosting Machine (LightGBM) models based on attachment anxiety, jealousy cognitions, social media surveillance, emotional dependency, and impulsivity.</p> <p><strong>Methods and Materials:</strong> The study employed a cross-sectional predictive correlational design using machine learning methodology. The statistical population consisted of emerging adults aged 18 to 29 years residing in major urban regions of Canada during 2025–2026. A total of 1,137 participants were selected through online and university-based recruitment procedures. Data were collected using the Relationship Stability Scale, Experiences in Close Relationships-Revised questionnaire, Multidimensional Jealousy Scale, Social Media Surveillance Scale, Emotional Dependency Questionnaire, and Barratt Impulsiveness Scale-Version 11. Data analysis was conducted using Python and machine learning libraries including Scikit-learn and LightGBM. The dataset was divided into training and testing subsets using an 80/20 split procedure. Model performance was evaluated using accuracy, precision, recall, F1-score, area under the curve (AUC), and log-loss indices. SHAP analysis was additionally used to examine predictor importance and model interpretability.</p> <p><strong>Findings:</strong> The findings demonstrated significant positive correlations among all study variables and romantic relationship instability. Emotional dependency showed the strongest correlation with relationship instability, followed by attachment anxiety, jealousy cognitions, social media surveillance, and impulsivity (p &lt; 0.01). The LightGBM model demonstrated high predictive performance with a testing accuracy of 0.88 and an AUC value of 0.91, indicating excellent discriminative ability. SHAP analysis revealed that emotional dependency and attachment anxiety were the most influential predictors of romantic instability. Hierarchical regression analysis further indicated that the predictor variables collectively explained 67% of the variance in relationship instability (R² = 0.67, p &lt; 0.001).</p> <p><strong>Conclusion:</strong> The findings suggest that romantic relationship instability among emerging adults is strongly associated with emotional insecurity, maladaptive jealousy cognitions, impulsive tendencies, and digitally mediated surveillance behaviors. The study further demonstrates that machine learning approaches such as LightGBM provide powerful tools for identifying complex nonlinear interaction patterns underlying romantic dysfunction. These findings highlight the importance of addressing attachment insecurity, emotional dependency, and unhealthy digital relational behaviors in prevention and intervention programs targeting young adults’ romantic well-being.</p> </td> </tr> </tbody> </table> https://journals.kmanpub.com/index.php/jayps/article/download/5444/9871