Wavelets in time-series analysis
Guy P. Nason, Rainer von Sachs · 2000
Abstract Reviewing the role of wavelets in statistical time-series analysis (TSA) appears to be quite an impossible task. For one thing, wavelets have become so popular that such a review could never be exhaustive. Another, more pertinent, reason is that there is no such thing as one statistical time-series analysis, as the very many different fields encompassed by TSA are, in fact, so different that the choice of a particular methodology must naturally vary from area to area. Examples for this are numerous: think about the fundamentally different goals of treating comparatively short correlated biomedical time-series to explain, for example, the impact of a set of explanatory variables on a response variable, or of analysing huge inhomogeneous datasets in sound, speech or electrical engineering, or, finally, building models for a better understanding and possibly prediction of financial time-series data.