I have got the power spectral density of a time series. Now I want to know the amount of noises present in the time series using MLE. Please help me out..
No products are associated with this question.
Im just guessing here, but since the least square fit of your time function (inverse fourier?) is a maximum likelihood estimator of your signal model (assuming gaussian noise), then the sum of residuals should be a measure for the amount of noise?
Isn't the average a max likelihood estimator? So can't you just subtract your signal from the average signal in the time domain? You can use conv() if you want a sliding average.