Add get_numeric_columns method in VMDataset to get list of all numeric columns

tests/test_dataset.py

@@ -66,6 +66,50 @@ def test_init_dataset_pandas_feature_columns(self):
         self.assertEqual(vm_dataset.feature_columns_categorical, [])
         self.assertEqual(vm_dataset.feature_columns, ["col1"])
 
+    def test_get_numeric_columns(self):
+        """
+        Test that get_numeric_columns returns all numeric columns in the dataset,
+        including target and extra columns, not just feature columns.
+        """
+        # Dataset: col1 (numeric), col2 (object/categorical), text_col (text),
+        # target (numeric)
+        df_with_text = pd.DataFrame(
+            {
+                "col1": [1, 2, 3],
+                "col2": ["a", "b", "c"],
+                "text_col": ["first review", "second review", "third review"],
+                "target": [0, 1, 0],
+            }
+        )
+        vm_dataset = DataFrameDataset(
+            raw_dataset=df_with_text,
+            target_column="target",
+            text_column="text_col",
+            feature_columns=["col1", "col2"],
+        )
+        numeric_cols = vm_dataset.get_numeric_columns()
+        self.assertEqual(set(numeric_cols), {"col1", "target"})
+        self.assertNotIn("col2", numeric_cols)
+        self.assertNotIn("text_col", numeric_cols)
+
+        # feature_columns_numeric only has feature columns that are numeric
+        self.assertEqual(vm_dataset.feature_columns_numeric, ["col1"])
+        # get_numeric_columns includes target as well
+        self.assertIn("target", numeric_cols)
+
+        # All-numeric dataset: get_numeric_columns returns all columns
+        df_numeric = pd.DataFrame({"a": [1, 2], "b": [3.0, 4.0], "c": [5, 6]})
+        vm_numeric = DataFrameDataset(raw_dataset=df_numeric)
+        self.assertEqual(
+            set(vm_numeric.get_numeric_columns()),
+            {"a", "b", "c"},
+        )
+
+        # No numeric columns (only object/categorical/text)
+        df_cat = pd.DataFrame({"x": ["a", "b"], "y": ["c", "d"]})
+        vm_cat = DataFrameDataset(raw_dataset=df_cat)
+        self.assertEqual(vm_cat.get_numeric_columns(), [])
+
     def test_dtype_preserved(self):
         """
         Test that dtype is preserved in DataFrameDataset.

validmind/vm_models/dataset/dataset.py

@@ -885,6 +885,21 @@ def df(self) -> pd.DataFrame:
         else:
             return as_df(self._df[columns]).copy()
 
+    def get_numeric_columns(self) -> List[str]:
+        """
+        Returns the names of all columns in the dataset that have numeric dtype.
+
+        Unlike feature_columns_numeric, this includes every column in the dataset
+        (target, extra columns, prediction columns, etc.) that is numeric.
+
+        Returns:
+            List[str]: The names of all numeric columns in the dataset.
+        """
+        dtypes = self._df.dtypes
+        return dtypes[
+            dtypes.apply(lambda x: pd.api.types.is_numeric_dtype(x))
+        ].index.tolist()
+
     @property
     def x(self) -> np.ndarray:
         """