Repository navigation
OpenML Parameter to Sklearn Parameter and vice versa #564
Description
Currently missing functions, I would like to add them and I think they would belong in this library
{{ message }}
Currently missing functions, I would like to add them and I think they would belong in this library
Activity
mfeurer commentedon Oct 5, 2018
Could you please be a bit more specific? I don't really understand this issue.
janvanrijn commentedon Oct 5, 2018
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_leafhyperparameter.The parameter name is
min_samples_leafThe scikit-learn name is
tree__min_samples_leafThe 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)
Adds flow.get_structure and flow.get_subflow (which are complements o…
mfeurer commentedon Dec 11, 2018
Added via #564 by @janvanrijn