0.9.0

[rcap107]

✨ Skrub version 0.9.0 has been released ✨

Main changes

• Scorers used by Data Ops can now take additional arguments (like sample weights). By @jeromedockes in allow passing additional arguments to scorers in dataops #1995
• The new methods .skb.find() and .skb.find_X_y have been added. .skb.find() can find a node by name (or by a callable predicate) in a DataOp. .skb.find_X_y() finds the nodes marked with .skb.mark_as_X() and .skb.mark_as_y(), and the cv splitter and split_kwargs passed to .skb.mark_as_X(), if they exist. By @jeromedockes in add methods to find a node in a dataop #2041
• The Cleaner now exposes a parse_numbers boolean parameter to control whether numeric-looking strings (e.g., ["1", "2", "3"]) are parsed to float32, and a cast_to_float32 parameter to downcast numeric columns to float32. By @Varshith-Yadav in [ENH]Refine Cleaner numeric_dtype modes #1910
• TableReport now accepts plot_distributions and compute_associations parameters (True, False, or "auto") to explicitly control whether distribution plots and pairwise associations are computed. By @JulietteBgl in ENH: Be more explicit about max_*_columns in TableReport #1907
• The diagrams displayed in notebooks for SkrubLearner, ParamSearch and OptunaParamSearch have been improved and now display the DataOp they contain. By @jeromedockes in show dataop in SkrubLearner estimator diagram #2024
• Added a metric parameter to fuzzy_join() and Joiner to configure the nearest-neighbor distance used for matching. By @sabasiddique1 in ENH - Add configurable distance metric for fuzzy join nearest‑neighbor matching #1861
• The has_dtype() selector has been added, which users to select columns by passing the dtype objects they want to match. By @kudos07 in ENH add has_dtype selector #2027
• ApplyToCols now has a exclude_cols parameter, to choose which columns should not be transformed. by @sabasiddique1 in ENH: add exclude_cols to ApplyToCols (#1873) #2039
• The number of associations shown by the TableReport has been increased. By @lisaleemcb in Increased the number of top associations to display #1992

New Contributors

• @JulietteBgl made their first contribution in ENH: Be more explicit about max_*_columns in TableReport #1907
• @parulnith made their first contribution in DOC fix deduplicate example rendering #2046
• @Varshith-Yadav made their first contribution in [ENH]Refine Cleaner numeric_dtype modes #1910

Full Changelog: 0.8.0...0.9.0

---

This discussion was created from the release 0.9.0.