CODEWITHSUNDEEP
matlabpca
user3541989
user3541989

Reputation: 21

How to PCA reduction in Matlab

I am new to Matlab and have some problems using built in packages for PCA reduction. I have 37 objects each represented by 161 dimensional vector, that means i have 161 x 37 matrix called P. I need to reduce vector dimension to 3. so that each object will be represented by 3 dimensional vector. I have tried something with princomp(P) but i don't know which output to take.

[COEFF,SCORE] = princomp(P); newData=SCORE(:,1:3);

I think newData are not the right vectors ?

Upvotes: 2

Views: 1118

Answers (1)

Rafael Monteiro
Rafael Monteiro

Reputation: 4549

You have to transpose your data because princomp expects observations on rows:

[COEFF,SCORE] = princomp(P.');
newData=SCORE(1:3.:).';

Alternatively you can use pca function to give you only the first 3 principal components:

[COEFF,SCORE] = pca(P.', 'NumComponents', 3)
newData=SCORE.';

Upvotes: 2

Related Questions