Here's an interesting little function I cooked up. With many experimental spectra, there is an underlying 'background' signal that distorts the actual useful signal. Sometimes you can remove this by taking background spectra from somewhere where the actual signal is absent and subtract this (usually an average of several of them) from the useful spectrum.

In some cases, this isn't available, but the background is easy to simulate as, for example, it may be a straight line 'slope'. hence a simple function can be 'fitted' and subtracted.

However, there are clearly some cases where neither of these methodologies are possible. This post gives a simple function that I thought up and have used successfully.