How to interpret PCA

Hi, I was trying a principal component analysis and I'd like to get some help

First, here is a table that shows measured concentrations of dopamine (DA), 3,4-hydroxyphenylacetic acid (DOPAC), and homovanillic acid (HVA) in mice urine after 2 hours of brain electric stimulus. The stimulus intensity were control in 3 mice, 100 μA in 4 mice, and 200 μA in 4 mice.

Using pca function, How can I interpret that which hormone is the most significant?

Also, I'd like to draw a scatter plot in different colors in each group (control, 100, 200)?

[Star Strider]

How are you using pca with your data?

[Jin Woo PARK]

I tried using pca function typing as

[coeff_2hours,score_2hours,latent_2hours,explained_2hours,mu_2hours]=pca(x2hours)

and here are the output results

coeff_2hours=

0.079673 0.528817 0.844988

0.418385 0.751662 -0.50986

0.904769 -0.39415 0.161362

score_2hours=

-5.04709 -8.41625 -2.15417

46.53202 6.334792 0.147732

-53.3088 1.198444 4.712148

-0.68473 4.310732 -3.02132

-30.1071 2.136703 2.991445

47.59406 -6.29931 3.21377

-25.753 -0.84886 -3.01205

3.155725 -2.33881 1.021516

20.26302 1.733989 -1.06366

-20.2205 -0.14942 -4.20901

17.57651 2.337988 1.373598

latent_2hours=

1000.621

18.99142

8.612187

explained_2hours=

97.31541

1.847012

0.837579

mu_2hours=

14.38647 27.68305 61.08517

[Image Analyst]

You forgot to attach x2hours. Please do so

save('answers.mat', 'x2hours');

Also, I don't know what you plotted but I'm not sure I see any point in using PCA since your data looks like a shotgun blast - totally uncorrelated.

Answer 1

[the cyclist]

@Jin Woo:

If you search this forum for cyclist and pca, you will find a few question/answers where I explain a lot about the interpretation of PCA output. Here are a couple you might want to look at:

• how to select components that show the most variance in PCA?
• how to apply PCA correctly?