A note on wavelet shrinkage in nonparametric regression models with ARFIMA errors

TL;DR

This paper introduces a wavelet shrinkage technique for nonparametric regression with ARFIMA errors, demonstrating improved performance over traditional thresholding methods through simulation studies.

Contribution

It proposes a novel wavelet-based shrinkage method tailored for models with ARFIMA errors, enhancing signal estimation accuracy.

Findings

The proposed method outperforms universal thresholding in simulations.

Monte Carlo experiments validate the effectiveness of the new approach.

Improved estimation accuracy in the presence of ARFIMA errors.

Abstract

In this paper we propose a shrinkage wavelet-based method to estimate the signal in a nonparametric regression model with Autoregressive Fractionally Integrated Moving Average (ARFIMA) errors. Monte Carlo experiments indicate that the proposed method is better than the universal thresholding rule which is widely used in data analysis via wavelet regression models.