Monotone Least Squares and Isotonic Quantiles

We consider bivariate observations such that, conditional on the , the are independent random variables with distribution functions , where is an unknown family of distribution functions. Under the sole assumption that is isotonic with respect to stochastic order, one can estimate in two ways: (i) For any fixed one estimates the antitonic function via nonparametric monotone least squares, replacing the responses with the indicators . (ii) For any fixed one estimates the isotonic quantile function via a nonparametric version of regression quantiles. We show that these two approaches are closely related, with (i) being more flexible than (ii). Then, under mild regularity conditions, we establish rates of convergence for the resulting estimators and , uniformly over and in certain rectangles as well as uniformly in or for a fixed .
View on arXiv