Does PCA funciton normalize the data while it generates the biplot?

3 次查看（过去 30 天）

Yaser Khojah 2019-5-7

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/460772-does-pca-funciton-normalize-the-data-while-it-generates-the-biplot

EF_Clu_Det.txt

Dear all,

I have 16 variables and I would like to use PCA to understand how they are correlated among each other’s. The first 8 variables (rates) have different units than the last 8 ones (times). After i used the code (as below) and created the biplot, I see the first 8 variables (rates) have an impact only. However, the rest of 8 variables (time) have small impact and I can’t see them on the graph. Does that happen because they are in different units? does PCA take care of variables when they are in different units? Is there a way to fix this? I'm positive these variables (time) should have a strong impact? Is there a way to fix this such as normalizing the data before running the code?

categories = ['r1';'r2';'r3';'r4';'r5';'r6';'r7';'r8';...
              't1';'t2';'t3';'t4';'t5';'t6';'t7';'t8'];
          
 [coefforth,score,~,~,explainedVar] = pca(EF_Clu_Det(:,1:16));
 
 biplot([coefforth(:,1) coefforth(:,2)],'Scores',[score(:,1) score(:,2)],'Varlabels',categories);