Extensions of smoothing via taut strings

TL;DR

This paper introduces extended smoothing methods using taut string algorithms for estimating various regression functions, including adaptive local total variation penalties, with proven consistency and practical algorithms.

Contribution

It extends smoothing via taut strings to generalized regression models, including binary and Poisson, with adaptive penalties and noniterative algorithms.

Findings

Developed noniterative algorithms related to taut string methods.

Extended framework to binary and Poisson regression.

Proved consistency of the proposed estimators.

Abstract

Suppose that we observe independent random pairs $(X_1,Y_1)$, $(X_2,Y_2)$, >..., $(X_n,Y_n)$. Our goal is to estimate regression functions such as the conditional mean or $\beta$--quantile of $Y$ given $X$, where $0<\beta <1$. In order to achieve this we minimize criteria such as, for instance, $$ \sum_{i=1}^n \rho(f(X_i) - Y_i) + \lambda \cdot \mathop TV\nolimits (f) $$ among all candidate functions $f$. Here $\rho$ is some convex function depending on the particular regression function we have in mind, $\mathop {\rm TV}\nolimits (f)$ stands for the total variation of $f$, and $\lambda >0$ is some tuning parameter. This framework is extended further to include binary or Poisson regression, and to include localized total variation penalties. The latter are needed to construct estimators adapting to inhomogeneous smoothness of $f$. For the general framework we develop noniterative algorithms for the solution of the minimization problems which are closely related to the taut string algorithm (cf. Davies and Kovac, 2001). Further we establish a connection between the present setting and monotone regression, extending previous work by Mammen and van de Geer (1997). The algorithmic considerations and numerical examples are complemented by two consistency results.