I have 2d data-set 100*100 for steady case. I have added 1% guassian noise. Can you please tell me how I can de-noise that by using singular spectrum analysis or any other method. Thanks!
Konstantin Zaytsev This is the file with noisy data and clean data.Thanks for your time. I will wait for your reply.
I used SVD on a noisy dataset. For example, how this can be done in the MATLAB: [U,S,V] = svd(data). In this case, S is a diagonal matrix. Then I left the first N elements of the S matrix, the rest set 0. After that, I got the composition A = U*S*V'. A matrices are presented on sheets "svdN". N is 2; 3; 15. Sheets "sko_N" show the deviation of values from the clean dataset. In cell A1 - the average deviation in dataset. When N = 3, the minimum deviation was obtained. Try another N = 4, for example.
- Badges
- Science topic