Dear colleagues,

Empirical mode decomposition families versus wavelet families (DWT, MODWT, EWT)

does exist a specific rule that allows us to use all decomposition components extracted using empirical modes decomposition method or empirical wavelet transform (EWT) as input feed ML models (same as what we do in the DWT pre-processing) instead of applying them as an individual prediction and summing all outcomes of each decomposition component? if possible please get me a reliable reference or participate in a discussion with each other!

most of the literature on forecasting problems uses empirical mode decomposition families and EWT using separately predicting each IMF and summing them as targets. various research believes there is no rule for this issue. please talk about your experiences!

Best regards

Mehdi