Time series analysis with the Hilbert–Huang transform

Time series data can be transformed to the frequency domain via the Hilbert–Huang transform. This transform has two major stages. The data are first deconstructed into a set of monocomponent intrinsic mode functions that are the basis states for the transform. A Hilbert spectral analysis is then performed on each of these basis states, yielding time dependent amplitude and frequency information. The process of extracting the intrinsic mode functions and performing Hilbert spectral analysis is described and applied to a set of examples that show the advantages and shortcomings of this transform.