New variables after computing Principal Components Analysis

After computing principal components analysis, what will represent my new variables to be used to build my model , is it the scores?

Thank You

Answer 1

[the cyclist]

Yes, the output score is the observed values of the variables in the principal coordinate axes.

I wrote a detailed answer to this question that is a tutorial on how to use MATLAB's pca() function. You might find it helpful.

[NCA]

Thanks a lot for taking the time out to respond and answer my question, I really appreciate. Am I right to only use the first few columns of my scores as shown by 'explained' ( example 80% explained variance) or do I need to use score multiplied by coeff (score*coeff) to build my model?

Thank You

[the cyclist]

You want to use score, not score*coeff.

score is equivalent to X*coeff; it's the original observations transformed to the PC coordinates.

Answer 2

[Image Analyst]

Attached is a demo where I get the PCs of an RGB true color image, plus another demo.