Dataframe run on task

[amueller]

This tries to solve #773.
Still probably needs some handling for sparse datasets.

[amueller]

@mfeurer can you explain the restriction on the third part of the subcomponent?

[mfeurer]

Please see my answer in #773

[Neeratyoy]

@amueller shall I try taking this PR up?

[Neeratyoy]

@mfeurer about 4/5 unit tests throw the following error when running on the latest sklearn version locally
openml.exceptions.OpenMLServerException: Problem validating uploaded description file - XML does not correspond to XSD schema

Also, I couldn't understand why for task.class_labels which is annotated as Optional[List[str]], it gives a flake error if I index it task.class_labels[0].
Flake error: Value of type "Optional[List[str]]" is not indexable

[mfeurer]

Also, I couldn't understand why for task.class_labels which is annotated as Optional[List[str]], it gives a flake error if I index it task.class_labels[0].

As it is typed optional it can actually be None, which would not be indexable. You probably need to use typing.cast to please mypy.

openml.exceptions.OpenMLServerException: Problem validating uploaded description file - XML does not correspond to XSD schema

Hm, apparently, the dummy imputor needs a docstring, otherwise the description uploaded to OpenML will be emptly.

[codecov-io]

Codecov Report

Merging #777 into develop will decrease coverage by 0.42%.
The diff coverage is 68.83%.

@@             Coverage Diff             @@
##           develop     #777      +/-   ##
===========================================
- Coverage    88.57%   88.15%   -0.43%     
===========================================
  Files           37       37              
  Lines         4324     4363      +39     
===========================================
+ Hits          3830     3846      +16     
- Misses         494      517      +23

Impacted Files	Coverage Δ
openml/__init__.py	100% <100%> (ø)	⬆️
openml/config.py	78.99% <51.61%> (-11.58%)	⬇️
openml/datasets/dataset.py	85.98% <80%> (-1.05%)	⬇️
openml/datasets/functions.py	95.07% <80%> (+0.95%)	⬆️
openml/_api_calls.py	87.93% <0%> (-2.59%)	⬇️
openml/runs/functions.py	82.83% <0%> (-0.28%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 371911f...ae9c312. Read the comment docs.

[mfeurer]

This looks good to me by now :) I have added a few more comments, and I think there are few comments still open.

[mfeurer]

Thanks for addressing my change requests, that all looks good now. There are a few old comments though, which are collapsed and you need to press 'show more to see them, that would be great if you could have a look at them. We can then ask @PGijsbers for another look and finally merge this :)

[PGijsbers]

Before I dive into the meat of this PR (which I will, with additional comments), I figured I would already add change requests/clarifications to get things moving :)

[PGijsbers]

I'll approve this PR so we can push a release before the workshop. Please open a new Github issue which links back to this PR so we have an easy reference to some further improvements that can be done in these modules/examples.