Calculate entropy of a thresholded correlation matrix
5 vues (au cours des 30 derniers jours)
Afficher commentaires plus anciens
Greetings All!
I am reasonably new to coding and would very much like your opinion and advice.
I have several time series for nodes of different networks. I want to create correlation matrices from selections of these time series and compare the entropy of different correlation matrices. I have the option of thresholding these matrices to get a top percentage (it is noisy data) and possibly even binarising these according to the threshold.
Can I simply use matlab's "entropy" function for a grey scaled image validly? https://uk.mathworks.com/help/images/ref/entropy.html.
Or could I use wentropy(x , 'shannon') https://uk.mathworks.com/help/wavelet/ref/wentropy.html which i belive is meant for wave signals
I understand that the entropy will vary according to the size of the matrix (my matrices vary from 61x61 to 213x213). Thus should I just code the normalised shannon entropy?
%%%%%%%%%
X_uni = unique(X);
X_uni_size = numel(X_uni);
n = numel(X)
P = zeros(X_uni_size,1);
for i = 1:X_uni_size
P(i) = sum(X == X_uni(i));
end
P = P ./ numel(X);
% Compute the Normalised Shannon's Entropy
H = -sum(P .* log2(P))/log2(P).*n
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
As an alternative I was thinking of just colapsing these matrices into a vector and using something like sample entropy.
Ideally I would want to use several types of complexity measure on these matrices so any suggestion is very welcome !
Best
Peter
0 commentaires
Réponses (0)
Voir également
Catégories
En savoir plus sur Denoising and Compression dans Help Center et File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!