OpenML Parameter to Sklearn Parameter and vice versa

[janvanrijn]

Currently missing functions, I would like to add them and I think they would belong in this library

[mfeurer]

Could you please be a bit more specific? I don't really understand this issue.

[janvanrijn]

Hyperparameters in our eco-system have generally three names. Take for example a pipeline with a decision tree (called: tree). Let's consider the min_samples_leaf hyperparameter.

The parameter name is min_samples_leaf
The scikit-learn name is tree__min_samples_leaf
The openml name is sklearn.trees.DecisionTree(1)_min_samples_leaf (or, one could argue that this hyperparameter can be uniquely identified by a flow id and a hyperparameter name)

I want to make a function that maps the openml hyperparameter name to a scikit-learn hyperparameter name (just realized the other way around is a little bit more complicated, because of version numbers)

[mfeurer]

Added via #564 by @janvanrijn