Matthias Feurer

Hey, I transferred this issue to the main OpenML issue tracker as this is not an issue of the PYthon API.

I agree that we should try to simplify this. I'm just posting the outcomes of our offline discussion for reference: If there was a machine readable version of the API...

After a discussion with @amueller I realized that this is a pretty bad explanation of what we would like to have. It should be a dictionary mapping from software to...

> It would be great if there was a way to get the parameter settings from a run in a way that is similar to how they are structured in...

> I saw get_structure but I did not understand the docstring. Maybe an example would be helpful. @janvanrijn > It would be really great if there was some documentation on...

> From what I can tell, @janvanrijn's contrib library is quite outdated, but I haven't checked in detail. I guess Jan hasn't adapted the library to the recent extension interface....

This basically sounds like adding in a layer of abstraction between the OpenML schema (`OpenMLFlow` class) and the scikit-learn object? Besides the refactor, I'd be afraid that we add a...

> Well, the OpenMLFlow class is not really the schema, right? No, it's not, but it tries to stay as close to the XML schema as possible. > The layer...

Great idea. Do you want to create a pull request for this repository and open an issue over at auto-sklearn? Thanks!

Hi @ArturDev42, Thank you very much for raising this issue. Yes, such lists are available, but very hidden and only accessible by using the API: * [Live server](https://www.openml.org/api/v1/estimationprocedure/list) * [Test...