In case of measured signals or data it is hard, if not impossible, to completely avoid all the possible sources of noise which can muddy any experimental measurement. But the presence of noise can perturb important characteristics of the data (e.g. the FWHM of a measured spectrum). Hence, it might be helpful to have some postprocessing tricks up one’s sleeve. Here we talk about just one such tool: the Savitzky-Golay filter, which smooths out local noise by performing a regression algorithm on groups of sampling points.
