Bayesian regression about scikit-learn

[fabianp]

Any progress on this ?

[vmichel]

Nope, but I will try it asap, in the next weeks.

[fabianp]

Cool. First learn mutt, then fix this :-)

[agramfort]

what's our status on this?

[vmichel]

Didn't work on this.

[amueller]

What do you think about my PR: #534 ?

[vmichel]

Seems good ! Thanks for cleaning this. Any way, I think that the Bayesian side of skl needs more love (and my TODO list is already long !)... Bayesian classification, different priors (using pyMC ?)

[amueller]

Thanks :)
Might need still more love ;)
I'm not actually using any Bayesian models, I just tried to use what I understood from Bishop's book to hopefully make this a little more clear ;)

[vmichel]

I'm not a Bayesian specialist too, so feel free to update this part of the SKL !

[amueller]

Do you still plan on working this? Otherwise I'd close the issue. If you have any more specific remarks, I'll gladly include them in my pull request.

[vmichel]

I will not work on this in the next weeks, so you can close the pull request. For me, due to the amount of work, Bayesian methods improvements in SKL should be dealt with in a dedicated sprint.