Comparison between empirical mode decomposition and local linear regression in presence of boundaries

Abstract

End points are areas of concern, as they are likely to influence data. The extension of data, or data prediction, is a risky procedure even for linear and stationary processes. The empirical mode decomposition (EMD) method is a useful analysis tool particularly when dealing with possibly non-stationary and nonlinear processes that characterize a time series. However, partial data information within boundaries is available because of the bounded support of the underlying time series. The same does not hold for nonlinear and non-stationary processes. This study compares two nonparametric methods, namely, EMD and local linear (LL) regression, in the presence of boundaries. A simulation study is conducted to assess the practical performance of the two methods.