Edoardo Abati

Thank you @aivanoved for working this! This is a great addition

Hi @TomBurdge I completely missed these last replies and I have started implementing something for this. Do you also have something implemented? We can collaborate on this and merge what...

> - A more sophisticated approach by @EdAbati: [draft-new-constructor](https://github.com/narwhals-dev/narwhals/compare/main...EdAbati:draft-new-constructor) (There is a fine line between 'more sophisticated' and 'overly complicated' 😅 I hope to finish the draft soon)

Uuuh this can be fun! I contributed to features related to the Array API compatibility lately. But I'll gladly also try to figure out this part of scikit-learn > it...

Some early thoughts: - what could be the cleanest way to select by row index? the pandas equivalent would be `X.iloc[shuffling_idx, col_idx]`, I guess polars also works with `X[shuffling_idx, col_idx]`,...

> Does implementing it interest you? Most of the time my answer to the question is "yes". My only problem is time 😅 I will create an issue in case...

Another useful feature could be something like polars' `DataFrame.replace_column`. Or do we have already the capability to overwrite a column without changing its position?

My daily struggle was again when the column name is an `int` 😅 If I'm not wrong, `with_columns` wants a string column name. `rename` seems to be happy with renaming...

FYI my changes are here: https://github.com/scikit-learn/scikit-learn/compare/main...EdAbati:scikit-learn:RESEARCH-narwhals Now the 3 functions linked above seem to work. Need to check if they were solved in main too and if I broke any...

Update: `cross_val_score` and `cross_validate` were fixed in `main`. The example with LogisticRegression mentioned in the original ticket works. Some test that use RandomForest fail, I am looking at those now....