Resumen de Discrimination of locally stationary time series
Andrés M. Alonso Fernández 
, Elizabeth Ann Maharaj
We propose a new method for the discrimination of locally stationary time series using wavelets. In particular we use time dependent wavelet variances as inputs into discriminant analysis. We will show that the wavelet variances are consistently estimated for locally stationary processes. Simulations studies show that our procedure performs very well, resulting in small training and hold-out sample classi cation errors. Applications to real data (seismic, electroencepha- logram and control chart data) show that our procedure performs as well and in some cases better than other classi cations methods.
