Addressing cluster-constant covariates in mixed effects models via likelihood-based boosting techniques

• DOI: 10.1371/journal.pone.0254178
• Publication Date: 2021-07-09T17:55:42Z
• Authors: Colin Griesbach; Andreas Groll; Elisabeth Bergherr
• Keywords: Boosting; Gradient boosting; Statistical Inference

Boosting techniques from the field of statistical learning have grown to be a popular tool for estimating and selecting predictor effects in various regression models can roughly separated two general approaches, namely gradient boosting likelihood-based boosting. An extensive framework has been proposed order fit generalized mixed based on boosting, however case cluster-constant covariates approaches tend mischoose variables selection step leading wrong estimates. We propose an improved algorithm linear models, where random are properly weighted, disentangled fixed updating scheme corrected correlations with improve quality estimates addition reduce computational effort. The method outperforms current state-of-the-art maximum likelihood inference which is shown via simulations data examples.