Piecewise monotone estimation in one-parameter exponential family

TL;DR

This paper extends piecewise monotone estimation techniques from normal means to general one-parameter exponential families, introducing an efficient algorithm and a data-driven change-point detection method with practical applications.

Contribution

It develops a new algorithm based on modified PAVA for exponential families and proposes an information criterion for regularization parameter selection.

Findings

Effective change-point detection in various data scenarios

Accurate spectrum estimation and causal inference results

Quantification of discretization errors in ODE solvers

Abstract

The problem of estimating a piecewise monotone sequence of normal means is called the nearly isotonic regression. For this problem, an efficient algorithm has been devised by modifying the pool adjacent violators algorithm (PAVA). In this study, we investigate estimation of a piecewise monotone parameter sequence for general one-parameter exponential families such as binomial, Poisson and chi-square. We develop an efficient algorithm based on the modified PAVA, which utilizes the duality between the natural and expectation parameters. We also provide a method for selecting the regularization parameter by using an information criterion. Simulation results demonstrate that the proposed method detects change-points in piecewise monotone parameter sequences in a data-driven manner. Applications to spectrum estimation, causal inference and discretization error quantification of ODE solvers are also presented.