While I write the monthly site statistics, I perform a Fourier analysis in the available data in Excel.

Excel has a built in tool for FFT, over 2^n samples strictly, and that analysis is done once; the results are pasted in the sheet and they don't refresh when new data is pasted. You have to run the tool again for the next power of two if you want new results.

Is there another way to do Fourier analysis over discrete data? Is it meaningful to expect different results for 90 and 100 samples?

More importantly, is there an algorithm that can be implemented with Excel formulas, so that the output is updated each time I insert a new sample?

Thanks in advance.

 

I wouldn't even attempt a DFT with excel formulas let alone an FFT.

But if you're brave enough to attempt a C to Excel translation then have a look here: http://en.literateprograms.org/Cooley-Tukey_FFT_algorithm_(C)

 

I was afraid that would be the answer. Thanks anyway. It's faster to port the data in Matlab for the next year than try to translate the code.