Calculating PCA

This should work fine. Feed in your 6x10 matrix of features, apply PCA, then reduce to 2 dimensions. This should work fine. If the results look good, that's great. If it doesn't work well (you won't know until you try), just use all 6 features for whatever analysis you want to do.

Why do you want to do this, by the way? If it is not for visualization, you might as well not reduce dimensionality with PCA.

" If it is not for visualization, you might as well not reduce dimensionality with PCA." what is your alternative ?

Actually what I want to do that, I have above figure values as features of a healthy people, and another same set of different values of unhealthy people. So total 10x2=20 features I will get. Each feature is 1x6 vector. This is called training.

Now I will take a new person data, and calculate his features. I will get 6 features i.e., 1x6 vector. Now I want to find weather this person (test) is related to Healthy people or unhealthy people.

Here I am thinking that PCA is useful.. Can you suggest me any other method to solve my problem?

Actually, you are trying to solve a classification problem with PCA. There are plenty of classification methods which you should try instead. Googling/Wikipediaing logistic regression, generative model for classification or k nearest neighbour classification might be what you want.

Can you please tell me the more appropriate classification technique to my problem.. Because my values are very much variant. Few are <1 and few are in thousands...