-
-
Notifications
You must be signed in to change notification settings - Fork 403
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Ensembles parallel #1116
Ensembles parallel #1116
Conversation
thanks for the help @PhilippPro |
|
- skipped stacking due to incoming refactoring - skipped prediction (left in branch) due to proposed parallelization of all prediction functions
365d047
to
4b1c274
Compare
all that should be done now (minus tutorials and new pr) |
there might still be some cleanup necessary. but i am out for the moment. |
@PhilippPro looks like the remaining errors are related to the multilabel stuff. help? |
ok. first have to look on my own stuff but will try it later. Am 12.08.2016 um 12:48 schrieb Zachary M. Jones:
|
thank you! |
if someone else wants to finish this up i would be grateful. i don't know the multilabel stuff (which is what i think is broken). |
I think there are more things that do not work. As you can see in the current travis build, the actual error occurs also in generatePartialDependence and MulticlassWrapper (and MultilabelDBRWrapper). I just solved the problem with the multilabel example in the commit before. |
ah ok i will take a look at it again then. thanks |
I can look at the multilabel part if everything else is ok. It's just disgusting to look at these errors, because you have to step deeply into the infrastructure of predictLearner, predictLearner2, etc. to find the problems and it takes a long time. |
# Conflicts: # R/generatePartialDependence.R
I fixed many variable missnamings and other stuff. But I can not get behind why the "MultilabelDBRWrapper" fails if it wasn't even touched! |
It depends on MultilabelBinaryRelevanceWrapper and this was touched. |
Oh I indeed oversaw this. I guess I know the issue then |
Will auto merge in 24 hours if no objections are raised. |
no, you can raise concerns that we are wasting your time. shout at me. i will try to review now. |
the indentation is wrong in several places. more importantly: |
True, documentation is totally missing. Any hint of where it would make sense to add it? Just add a note for each wrapper that it can be parallelized with the specific tag? indentation... 🙄 I will take care of it |
well, i think it needs to go here: BUT: I think we should also have this in R. makes it much easier to review stuff like this, whether it is "complete". and complete documentation about behavior of options should always be available in R. but we dont want to copy-paste docs to 2 places. I would suggest: for completeness we can add a sentence and a link to the wrapper that is affected by your change here. what do you think? |
So. Documentations is done, tests are running. Should be merged soon. |
@berndbischl ping |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It would be good to extend the tests to check the predictions as well.
We agreed that we do not need to check the result of the predictions as the |
not really. i detected many bugs with this test, that were not pm bugs. but problems in mlr. |
this is a clone of #615 in a branch in this repo so that @PhilippPro can help me fix the multilabel breakage