Two-stage variational mode decomposition approach to enhance the estimates of variance function

Abstract

A two-stage variational mode decomposition approach is explored in this article to estimate the variance function in a nonparametric heteroscedastic fixed design regression model. This estimate improves the results obtained by using variational mode decomposition in the case of smooth regression function. We noticed that the amount of improvement is depending on the smoothness of the variance function. Simulations and comparisons with the earlier methods are carried out to evaluate the performance of the proposed method.

Keywords:

• Heteroscedasticity
• Difference-based initial estimates
• Variational mode decomposition
• Two-stage VMD
• Smooth regression function