How do you determine if it’s better to standardize your data matrix when you perform an analysis of the main components in R?

Question

I am trying to analyze the main components in R. I believe there are two ways to do this. One of them does the analysis of the main components immediately, another way is to first standardize the matrix using s = scale (m), and then apply the analysis of the main components.
How to find out which result is better? What values ​​should I look at. I have already managed to find the eigenvalues ​​and eigenvectors - the variance fraction for each eigenvector, using both methods.

I noticed that the proportion of variance for the first pca without standardization was more important. Is there any reason? Isn't that always the case?

Finally, if I have to predict a variable, i.e. weight should i reset the variable i.e. the weight from my data matrix when I perform the analysis of the main components?

Answer 1

Are your variables measured on a global scale? If so, do not scale. If not, then probably a good idea to scale.

If you are trying to predict the value of another variable, the PCA is probably not the right tool. Perhaps you should take a look at the regression model.