How to extract the variable IDs after dimensionality reduction via pca?

Question

Bank Yin 2014-7-8

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/141014-how-to-extract-the-variable-ids-after-dimensionality-reduction-via-pca

评论： Bank Yin 2014-7-8

I have a dataset X(voxel*protein), it's size is 3694200*10. I used pca to reduce dimensionality and got a matrix of size 3694200*4. The question is that I need to know which 4 proteins are remained. How can I get the original column IDs(0-9) of the 4 proteins?

My code: [COEEF,SCORE,LATENT] = pca(X); cumVar = cumsum(LATENT)./sum(LATENT); reducedData = SCORE(:,1:4);

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

the cyclist 2014-7-8

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/141014-how-to-extract-the-variable-ids-after-dimensionality-reduction-via-pca#answer_144237

PCA doesn't select a subset of variable. (Specifically, in your case, PCA does not select 4 out of your 10 variables).

Instead, PCA is identifying linear combinations of your original variables that explain the overall variation. If a small number of these linear combinations capture most of the variation, then it makes sense to limit to those combinations. The first output variable tells you the coefficients of those linear combinations.