Semi-non parametric smooth isotonic regression

We propose a new method for smooth isotonic regression analysis. Unlike most existing methods for isotonic regression, the proposed method is akin to parametric regression without order restriction. To account for smoothness and isotonicity simultaneously, we exploit the flexible class of semi-non parametric densities to model isotonic regression functions. Under this framework, the full range of inference techniques for parametric regression models become applicable for model estimation and model validation in isotonic regression.