On Thu, Oct 04, 2012 at 10:55:03PM +0200, Nelle Varoquaux wrote: > This problem does not occur with the r version of the lars lasso.
I don't know R. Could you send me R code to reproduce the behavior under R. I've been looking at the problem more precisely, and in our implementation, the gram matrix of the active set becomes very very ill-conditioned (1e-15 eigenval). Choleski of the gram matrix blows up. I wonder how they avoid that problem in R. The naive implementation of Alejandro avoids it by using linalg.solve, but that really defeats the purpose of lars, which is really to iteratively compute only updates. Gaël ------------------------------------------------------------------------------ Don't let slow site performance ruin your business. Deploy New Relic APM Deploy New Relic app performance management and know exactly what is happening inside your Ruby, Python, PHP, Java, and .NET app Try New Relic at no cost today and get our sweet Data Nerd shirt too! http://p.sf.net/sfu/newrelic-dev2dev _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
