Authors: Alsaidi M. Altaher and Mohd Tahir Ismail
Polynomial wavelet regression (PWR) is usually considered as treatment for boundary problem present in wavelet regression. Practical implementations of this combined estimator might be faced by the presence of missing data in the response variable. Such missing observations can pose serious problems, affect the bias and finally misleading conclusions might be drawn. This paper introduces Bootstrap-Iterative Polynomial Imputation (BIPI) as a simple, an efficient method to replace the missing values in (PWR). Simulation study is conducted to assess the numerical achievement of (BIPI) with comparison with: mean-median substitutions, non-parametric bootstrap and EM algorithm.
Keywords: Missing data; wavelet regression; simulation