Hi,
I was looking to the implementation of lasso_stability_path
and how can be clearly seen here
http://scikit-learn.org/0.13/auto_examples/linear_model/plot_sparse_recovery.html
the alphas values returned by the algorithm are normalize.
While this is closer to what is reported in the paper I think that to make
the algorithm more flexible the returned value should be alphas (and not
alphas/max(alphas))
This would make the algorithm also more similar to the others that use the
lars path.
This is just a my personal point of view so maybe there are deeper reasons
that I have not considered.
Best,
Luca
------------------------------------------------------------------------------
Slashdot TV.
Video for Nerds. Stuff that matters.
http://tv.slashdot.org/
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
