Bayesian selection of primary resolution and wavelet bias functions for wavelet regression (Q626208)

A shrinkage rule for wavelet regression is proposed where the primary resolution \(m\) and the set of basis functions with non-zero coefficients are selected via a Bayesian approach. Non-informative (flat) priors for \(m\) and the number \(s\) of ``significant'' basis functions are given. The basis functions are rearranged in order of their ``importance'' and \(s\) ``most important'' functions are selected. The performance of the proposed algorithm is investigated via simulations.