December 20, 2013 gully; Austin, TX
This is bivariate practice from the astroML textbook chapter 3.0 figures The original code is located at:
http://www.astroml.org/book_figures/chapter3/fig_robust_pca.html
The file fig_robust_pca.py is the original.
The file fig_robust_pca_gull.py is an edit by gully.
The key idea was that I explored the dependence of the robust and non-robust fitting techniques by trying contamination fraction from 0.001 to 0.70. The original example only looked at 5% and 15% outliers.
Hope you enjoyed the example, and that you can get it to run.
Best wishes, -gully