Robust logistic zero-sum regression for microbiome compositional data

Abstract We introduce the Robust Logistic Zero-Sum Regression (RobLZS) estimator, which can be used for a two-class problem with high-dimensional compositional covariates. Since log-contrast model is employed, estimator able to do feature selection among parts. The proposed method attains robustness by minimizing trimmed sum of deviances. A comparison performance RobLZS non-robust counterpart and other sparse logistic regression estimators conducted via Monte Carlo simulation studies. Two microbiome data applications are considered investigate stability presence outliers. available as an R package that downloaded at https://github.com/giannamonti/RobZS .