Using fourier analysis for time series prediction

Question

For data that is known to have seasonal, or daily patterns I\'d like to use fourier analysis be used to make predictions. After running fft on time series data, I obtain coeffic

Answer 1

When you run an FFT on time series data, you transform it into the frequency domain. The coefficients multiply the terms in the series (sines and cosines or complex exponentials), each with a different frequency.

Extrapolation is always a dangerous thing, but you're welcome to try it. You're using past information to predict the future when you do this: "Predict tomorrow's weather by looking at today." Just be aware of the risks.

I'd recommend reading "Black Swan".