mathurinm

@hoodaty can you also edit the first message of this PR to link to the related issue? You need to add a unit test for this, again a comparison against...

closing as a duplicate of the merged #258

@tomaszkacprzak we can start by supporting sparse matrices for the GroupBCD solver. Do you think you could send a PR doing that ? You'd need to implement `construct_grad_sparse` and `bcd_epoch_sparse`...

I think it's a popular enough model such that the visibility benefits of implementing its class overweighs the cost of maintenance: I am +1 on the `GroupLasso` estimator

@tomaszkacprzak did you try our implementation of `positive` for the Group Lasso ? ANd would you be interested in implementing sparse X support for group ?

This should be fixed now @tomaszkacprzak, thanks @Badr-MOUFAD

Fixed by #228

@PABannier thanks a lot ! I tried it and it really shines on the data by @sehoff Caveat: the stopping criteria are not the same. Monitoring the primal decrease is...

@sehoff hard to tell without the data but consider increasing max_iter, probably the solver is not very far from convergence.You can run it in verbose mode to see if you...

@sehoff support for weights (inifnite ones too) is here. Let us know how it works !