Principal Component Analysis

Hi there,

I want to find significant chemical compound from 7*64 matrix of data, so that i can do classification using only those significant only. I normalized it first using zscore (range -2 to 2.5), then i apply PCA to reduce the data. I got coeff (64*64), score(7*64), latent(64*1), explained variance (bargragph of 5components), eigenvalues(64*64), eigenvectors(64*64), etc. however, i still dont know what is my significant chemical compound? is that the score value?or, how can i find which one in the original data (7*64) that has been chosen as the last output as principal component ?

Many thanks

Laila

Answer 1

[Richard Willey]

The Statistics Toolbox product page has a good demo titled "Partial Least Squares Regression and Principal Components Regression"

http://www.mathworks.com/products/statistics/demos.html?file=/products/demos/shipping/stats/plspcrdemo.html

You should be able to modify this example pretty easily...