ESTIMATION OF FUNCTIONAL-COEFFICIENT AUTOREGRESSIVE MODELS BY WAVELET METHODS

Pedro A. Morettin (Brazil) and Chang Chiann (Brazil)

Received December 20, 2007

Abstract

In this paper we use wavelet methods for the estimation of functional coefficient autoregressive models. Specifically, we propose linear and nonlinear wavelet estimators. These are time-domain estimators in the sense that they involve the ordinary least squares method. We compare the proposed estimators with other estimators through simulation studies and also present applications to real data sets.

Keywords and phrases: autoregressive models, design-adapted wavelets, functional coefficients, least squares, warped wavelets, wavelets.