Principal Component Analysis

[msiamj]

Dear all,


I would like to use the principal component analysis (PCA) in my studies; however this is new for me. When I go to proc -> make principal component in Eviews 6, there is a box with several choices which I not sure which to use.

Under “Scaling” there are four options: Normalize loadings, Normalize scores, Symmetric weights, and User loading weights, where under another column which is “covariance specification” there are 2 options: Correlation and covariance.

May I know what exactly the options are about so that I can decide which to choose?

Thank you!

[EViews Glenn]

The settings correspond to what properties you want your components to have...

Normalize loadings - means that your scores will have variances equal to the estimated eigenvalues
Normalize scores - means that your scores will have unit variances

The latter corresponds to whether you want the eigenvalue decomposition computed using the covariance or the correlation matrix of the data.

The default settings are probably the most commonly used, though the choice of correlation (default) versus covariance is generally a matter of application.

[msiamj]

Hi,

Thank you so much for the explanation.

May I ask another question? If I want to compute the new set of data using PCA, what I need to do is just insert the data, then go to proc > make principal component? When I try to read more about PCA, all the tutorials show many steps in order to get the data, but they are using software like SPSS and SAS. But for EVIEWS I can just get it by clicking "make principal component" right?

Appreciate your reply. This really helps.

[EViews Glenn]

It's hard to answer definitively without seeing the other tutorials, but once you have the group in EViews you should have your source data set up. Clicking on Make Principal Components... will compute the relevant correlation or covariance matrix, compute the eigenvalue decomposition, then make the scores corresponding to your settings...

[msiamj]

Hi,

I will try it, thanks a lot!