281
v1v2v3 (latest)

Least Squares estimation of two ordered monotone regression curves

Abstract

In this paper, we consider the problem of finding the Least Squares estimators of two isotonic regression curves g1g^\circ_1 and g2g^\circ_2 under the additional constraint that they are ordered; e.g., g1g2g^\circ_1 \le g^\circ_2. Given two sets of nn data points y1,...,yny_1, ..., y_n and z1,>...,znz_1, >...,z_n observed at (the same) design points, the estimates of the true curves are obtained by minimizing the weighted Least Squares criterion L2(a,b)=j=1n(yjaj)2w1,j+j=1n(zjbj)2w2,jL_2(a, b) = \sum_{j=1}^n (y_j - a_j)^2 w_{1,j}+ \sum_{j=1}^n (z_j - b_j)^2 w_{2,j} over the class of pairs of vectors $(a, b) \in \mathbb{R}^n \times \mathbb{R}^n $ such that $a_1 \le a_2 \le ...\le a_n $, $b_1 \le b_2 \le ...\le b_n $, and aibi,i=1,...,na_i \le b_i, i=1, ...,n. The characterization of the estimators is established. To compute these estimators, we use an iterative projected subgradient algorithm, where the projection is performed with a "generalized" pool-adjacent-violaters algorithm (PAVA), a byproduct of this work. Then, we apply the estimation method to real data from mechanical engineering.

View on arXiv
Comments on this paper