Correlation (cor) vs. Covariance (cov)

8 次查看（过去 30 天）
Newuser
Newuser 2011-5-1
Hi all,
does someone know why to compute a loop with the cor function takes way longer than doing it with the cov function?
I've the followin loop
covC = cell(szX(1),1);
CovM = zeros(szX(1),szX(2));
for n = ws1+1:szX(1)
covC{n} = cov(((logr(n-ws1:n,:))));
CovM(n,:)= (covC{n}(1,:));
end
the matrix is ca. (1580x32)... if u run the posted loop it takes just a few seconds. However If I want to compute the correlation (and substitute cov with cor) it takes about 5 minutes
corC = cell(szX(1),1);
CorM = zeros(szX(1),szX(2));
for n = ws1+1:szX(1)
corC{n} = corr(((logr(n-ws1:n,:))));
CorM(n,:)= (corC{n}(1,:));
end
. I want to do it also with a 3000x1200 matrix
can someone suggest me a quicker prodecure to achive my goal?
thaks for ur hepl
  2 个评论
Newuser
Newuser 2011-5-1
Problem solved by using corrcoef
Oleg Komarov
Oleg Komarov 2011-5-1
You can use the compiler to verify where are the bottlenecks in corr as compared to corrcoeff.

请先登录，再进行评论。

请先登录，再回答此问题。

回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Correlation and Convolution 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by