Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
April 12, 2026PsychometrikaOpen Access

Mixed-Effects Xgboost With Group-Aware Permutation Importance and Cross-Validation for Multilevel Cross-Classified Continuous Outcomes

View Full Paper
Ask AI
Bookmark
Share

Authors

SCSun‐Joo ChoSMSophia Mueller

Discussion

Loading...

Member takes

Overview

Demonstrates improved prediction accuracy in multilevel data, suggesting enhanced modeling for complex relationships.

Key Points

  • The research aims to develop a mixed-effects framework using XGBoost for analyzing multilevel cross-classified continuous outcomes.
  • Implemented LMM-XGBoost by embedding XGBoost within a linear mixed model.
  • Developed iterative estimation procedures and group-aware permutation importance.
  • Applied combined-group cross-validation for hyperparameter tuning and importance estimation.
  • Conducted a simulation study and an empirical application using the Add Health study.
  • LMM-XGBoost showed lower out-of-fold prediction error compared to standard models.
  • Achieved more accurate recovery of variable importance rankings.
  • Combined-group cross-validation yielded less biased error estimates.
  • Identified important factors related to adolescent depressive symptoms effectively.

Cite This Study

Cho et al. (2026) studied this question.

synapsesocial.com/papers/69db36e64fe01fead37c4dd6https://doi.org/10.1017/psy.2026.10108
View Full Paper
Ask AI
Bookmark
Share