Hi @DatenBiene ,

let me know if you still plan to work on this. Thanks!

Hi Roman @rth, I can work on this project now and will invest some time on it in the next weeks. Could you indicate me how the more helpful I could be on this and what kind of deliverable you expect? (Benchmarking with the R implementation? On how many datasets? Are there any rules I need to follow?)

Many thanks!

Great @DatenBiene ! There are two main points to start,

1. Address my review comments from the parent PR [MRG] Add quantile regression #9978 (review) and in particular remove L1 regularization to start with a simple approach.
2. try this implementation on a couple of classical datasets (say load_diabetes, fetch_california_housing + some regression dataset with known outlier issues) and check that a) it does converge, b) compare the runtime e.g. with respect to Ridge.

You should also resolve conflicts in doc/modules/classes.rst (click on "Resolve conflicts").

Benchmarking against a well tested R implementation (listed in the parent issue) would also be great.

Great ! I'll keep you inform of my improvement/difficulties. Thank you

Hi @DatenBiene are you still working on this? Thanks

Solved in #9978.