Recommended reading from Statistics, Data Mining, and Machine Learning
in Astronomy ("SDMML") and Numerical Recipes ("NR"):
SDMML: Chapter 7
Example of PCA: this paper on galaxy classification
Try the simple 2-d PCA code in ~bromley/courses/ap7730/examples/pca2d.py. It runs PCA on pairs of x-y data, where {x_i,y_i} are uncorrelated. Modify so that x and y are correlated. Interpret the results.
Try out the simple "n"-d PCA code in ~bromley/courses/ap7730/examples/pcand.py. It runs PCA on data "vectors" x_i {x_1i,x_2i,...} where each component x_j of the i-th sample represents on of n "simultaneous" measurements. An example is that x_i might represent n channels in an analyzer or frequencies in a spectrum.
As in lecture, work on the "digits" problem.... ~bromley/courses/ap7730/examples/digits.py. It runs PCA on data