Mason Bretan

Technology, Art, and Science
   Home      Principal Component Analysis
First short lecture on PCA
1) What properties do good features exhibit
2) How can we use PCA do identify these good features?
1) Download the normalizedObservations.mat and use princomp in Matlab to get PCA coefficients, try it using the cov and pcacov functions
2) Download the observations.mat and normalize the features so that each feature has similar means and variances (my normalized observations aren't perfect, a few of the features have a slightly bigger variances so alternative norm functions can be tried)