Evolutionary Spectra for Exploratory Data Analysis
This paper gives a brief introduction to windowed Fourier analysis also known as evolutionary spectral analysis (ESA). ESA has largely fallen into a back- water because the type of analysis it was intended to perform is now usually done with wavelets. ESA is not identical to wavelet analysis and so could be a worthwhile analysis to routinely perform along side other more estab- lished methods. The paper presents some example time series of tree ring sequences and stock market volatilities and shows how evolutionary spectral analysis can be useful in the exploratory data analysis phase of analysing a time series.