Voxelwise glm residuals/R2 (cleaned up pull request)

[Gilles86]

This is a cleaner version of #173

[bthirion]

Thx for taking care of his.
My only real comment is that I agree to create a dedicated method, but we should avoid the use of setattr_on_read. The reason is that it makes it hard to understand the state of the objects when an error is raised, hence the code is hard to debug and errors are hard to reproduce. I think that recomputing the residuals is fine, as it is unlikely that people will do it many times.

Please also beware there the travis failure.

[Gilles86]

OK. Well I think I agree with you and I'm happy to make it a function instead of a "pseudo-attribute". Let's do that?

[bthirion]

Indeed, thx !

[jdkent]

I still would very much like voxelwise residuals, could I take up/finish this pull request?

[kchawla-pi]

Sure. Open up a new PR based on this branch.

…

On Fri, Nov 15, 2019, 19:49 James Kent ***@***.***> wrote: I still would very much like voxelwise residuals, could I take up/finish this pull request? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#298>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB6SXRBC326RU2TVPP35ZSDQT3OJZANCNFSM4GJDFYIQ> .

[bthirion]

Thx !