Re: [Scikit-learn-general] Lasso with alpha=0

On Sat, Feb 11, 2012 at 03:22:46PM +0100, Andreas wrote: > So we raise a ValueError in CD if alpha=0 A warning. > and remove this case from the test? Yes. Gaël -- Virtualization & Cloud Management Using Capacity Planni

Re: [Scikit-learn-general] Lasso with alpha=0

On 02/11/2012 02:30 PM, Alexandre Gramfort wrote: >> In "test_lasso_lars_vs_lasso_cd", ``lars_path(X, y, method='lasso') `` >> is called and returns some alphas. The last one is exactly zero. >> I'm not sure if this is supposed to happen. >> > yes it should be the case. And the solution for a

Re: [Scikit-learn-general] Lasso with alpha=0

> In "test_lasso_lars_vs_lasso_cd", ``lars_path(X, y, method='lasso') `` > is called and returns some alphas. The last one is exactly zero. > I'm not sure if this is supposed to happen. yes it should be the case. And the solution for alpha = 0 should match the OLS solution (at least with n_samples

[Scikit-learn-general] Lasso with alpha=0

Hi everybody. While I was looking through the test error messages, I came across some behavior in the linear_models that I don't understand. Maybe someone can enlighten me. In "test_lasso_lars_vs_lasso_cd", ``lars_path(X, y, method='lasso') `` is called and returns some alphas. The last one is exa