Christian Lorentzen

All 💚 with all 🌱. Thanks for your strong help @ogrisel.

@Micky774 Do you want to give it a second look and an approval? In that case I'll merge.

How to interpret an approval in combination with a "+0"? 😄 My motivation is to close #19615 and I think this PR does it.

The impact is as follows: - auto: see lbfgs, sag, cholesky, sparse_cg - svd: `tol` has no impact - cholesky: `tol` has no impact - lsqr: `tol` is set as...

With accuracy, I meant `max_abs_diff`, i.e. the maximum of relative differences of coefficients wrt the solution by svd. It shows that tol=1e-4 comes much closer to the correct values. Only...

@hannes Thanks for your prompt reply. Is there anything someone like me (newbie but big fan on duckdb) could help with?

If it's within reach for a non-expert and you can provide some guidance, I can give it a try. Some pointer would certainly help.

🤔 https://arrow.apache.org/blog/2019/09/05/faster-strings-cpp-parquet/ > One of the largest and most complex optimizations involves encoding and decoding Parquet files’ internal dictionary-encoded data streams to and from Arrow’s in-memory dictionary-encoded DictionaryArray representation.

With duckdb 0.5, this got a little better (~factor of 2): ```bash % ls -lh df* ... 9.6K Sep 17 19:09 df_arrow.parquet ... 204K Sep 17 19:09 df_duckdb.parquet ```

@elephaint already provided the answer. To make it even clearer: - As training loss: The Tweedie deviance is a function of labels (y) and raw scores. Take the Tweedie deviance,...