Distributional (Single) Index Models

A Distributional (Single) Index Model (DIM) is a semiparametric model for distributional regression, that is, estimation of conditional distributions given covariates. The method combination classical single-index models the mean response covariates, and isotonic regression. index parametric, whereas are estimated nonparametrically under stochastic ordering constraint. We show consistency our estimators apply them to highly challenging dataset on length stay (LoS) patients in intensive care units. use provide skillful calibrated probabilistic predictions LoS individual patients, which outperform available methods literature.