[FIX] Fix three robustness gaps in OpenMLDataset.__repr__, get_data(), and create_dataset

openml/datasets/dataset.py

@@ -1,6 +1,7 @@
 # License: BSD 3-Clause
 from __future__ import annotations
 
+import contextlib
 import gzip
 import logging
 import os
@@ -295,13 +296,6 @@ def id(self) -> int | None:
 
     def _get_repr_body_fields(self) -> Sequence[tuple[str, str | int | None]]:
         """Collect all information to display in the __repr__ body."""
-        # Obtain number of features in accordance with lazy loading.
-        n_features: int | None = None
-        if self._qualities is not None and self._qualities["NumberOfFeatures"] is not None:
-            n_features = int(self._qualities["NumberOfFeatures"])
-        elif self._features is not None:
-            n_features = len(self._features)
-
         fields: dict[str, int | str | None] = {
             "Name": self.name,
             "Version": self.version,
@@ -312,14 +306,23 @@ def _get_repr_body_fields(self) -> Sequence[tuple[str, str | int | None]]:
             "Pickle file": (
                 str(self.data_pickle_file) if self.data_pickle_file is not None else None
             ),
-            "# of features": n_features,
         }
+
+        # We don't consider these fields critical to display, so we use the internal
+        # attribute to avoid downloading extra files just to generate the repr.
+        try:
+            fields["# of features"] = int(self._qualities["NumberOfFeatures"])  # type: ignore[index]
+        except (TypeError, KeyError, ValueError):
+            if self._features is not None:
+                fields["# of features"] = len(self._features)
+
+        with contextlib.suppress(TypeError, KeyError, ValueError):
+            fields["# of instances"] = int(self._qualities["NumberOfInstances"])  # type: ignore[index]
+
         if self.upload_date is not None:
             fields["Upload Date"] = self.upload_date.replace("T", " ")
         if self.dataset_id is not None:
             fields["OpenML URL"] = self.openml_url
-        if self._qualities is not None and self._qualities["NumberOfInstances"] is not None:
-            fields["# of instances"] = int(self._qualities["NumberOfInstances"])
 
         # determines the order in which the information will be printed
         order = [
@@ -722,7 +725,7 @@ def valid_category(cat: Any) -> bool:
         raw_cat = pd.Categorical(col, ordered=True, categories=filtered_categories)
         return pd.Series(raw_cat, index=series.index, name=series.name)
 
-    def get_data(  # noqa: C901
+    def get_data(  # noqa: C901, PLR0912
         self,
         target: list[str] | str | None = None,
         include_row_id: bool = False,  # noqa: FBT002
@@ -794,6 +797,22 @@ def get_data(  # noqa: C901
             )
 
         target_name = target_names[0]
+        available_columns = list(data.columns)
+        if target_name in to_exclude:
+            msg = (
+                f"Target column '{target_name}' was removed because it is listed as a "
+                f"row_id or ignore attribute. "
+                f"Available columns after filtering: {available_columns}"
+            )
+            raise ValueError(msg)
+
+        if target_name not in data.columns:
+            msg = (
+                f"Target column '{target_name}' does not exist in this dataset. "
+                f"Available columns: {available_columns}"
+            )
+            raise ValueError(msg)
+
         x = data.drop(columns=[target_name])
         y = data[target_name].squeeze()
 

openml/datasets/functions.py

@@ -612,8 +612,7 @@ def create_dataset(  # noqa: C901, PLR0912, PLR0915
     # Edit: Found it could also be like ["True", "False"]
     attributes: list[tuple[str, str | list[str]]] | dict[str, str | list[str]] | Literal["auto"],
     data: pd.DataFrame | np.ndarray | scipy.sparse.coo_matrix,
-    # TODO(eddiebergman): Function requires `default_target_attribute` exist but API allows None
-    default_target_attribute: str,
+    default_target_attribute: str | None,
     ignore_attribute: str | list[str] | None,
     citation: str,
     row_id_attribute: str | None = None,
@@ -657,7 +656,8 @@ def create_dataset(  # noqa: C901, PLR0912, PLR0915
         passing ``attributes='auto'``.
         The target feature is indicated as meta-data of the dataset.
     default_target_attribute : str
-        The default target attribute, if it exists.
+        The default target attribute, if it exists. Use ``None`` for unsupervised datasets
+        (e.g. clustering, anomaly detection) where no natural target column exists.
         Can have multiple values, comma separated.
     ignore_attribute : str | list
         Attributes that should be excluded in modelling,