Automatic Singular Spectrum Analysis for Time-Series Decomposition

An automatic Singular Spectrum Analysis based methodology is proposed to decompose and reconstruct time-series. We suggest a clustering based procedure to identify the main dynamics of the input signal, by computing a subset of orthogonal basis using a power spectrum criterion. The subset of basis are represented by the Discrete Fourier Transform to infer basis vectors encoding similar data structures. Thus, it is possible to highlight hidden components into the signal. Our approach is tested over some synthetic and real-world datasets, showing that our algorithm is a good tool to decompose time-series.