Riccardo Cappuzzo

> an important intermediate step would be avoiding direct conversions of full dataframes between pandas and polars because those require pyarrow. in particular since the addition of pearson correlation we...

> ... > What do the others think? I agree on all points here and I would also prefer having both numbers visible at the same time. > Adding the...

> By your remarks, I see that the switching views is not interesting. I propose a static version where the percentage and the value are represented at the same time...

> * how do I set `max_plot_columns` to always plot? I would expect setting it to None to mean "no maximum" but it still uses the default 30 hah, that's...

> * BUG: setting `enable_subsampling` to "force" will also subsample even during `predict`; we never want subsampling in that case (`y_score` would have a different shape than `y_true`) For this...

To fully close this issue, setting `max_plot_columns=None` and `max_association_columns=None` should result in plotting/computing the associations for all columns. This should be done in the table report, where summarization is done.

From IRL conversation: this is mostly fine, but it should be summarized more, and links should be put into see also blocks.

Hi @honghanhh, thanks for the PR. I think it's already in a good shape, but I am not fully convinced by this implementation. I think it would have been better...

> @rcap107 I see your point and agree. The current implementation doesn’t let you drop columns when `drop_null_fraction < 1.0`. In this case, it makes sense to have option so...

> Maybe something to leave up for the sprint? yes that was the plan