refactor(python): Deprecate pl.get_dummies

py-polars/polars/internals/functions.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import contextlib
+import warnings
 from datetime import date, datetime, timedelta
 from typing import TYPE_CHECKING, Iterable, Sequence, overload
 
@@ -44,6 +45,9 @@ def get_dummies(
     """
     Convert categorical variables into dummy/indicator variables.
 
+    .. deprecated:: 0.16.8
+        `pl.get_dummies(df)` has been deprecated; use `df.to_dummies()`
+
     Parameters
     ----------
     df
@@ -75,6 +79,11 @@ def get_dummies(
     └───────┴───────┴───────┴───────┴───────┴───────┘
 
     """
+    warnings.warn(
+        "`pl.get_dummies(df)` has been deprecated; use `df.to_dummies()`",
+        category=DeprecationWarning,
+        stacklevel=2,
+    )
     return df.to_dummies(columns=columns, separator=separator)
 
 

py-polars/tests/unit/datatypes/test_categorical.py

@@ -64,7 +64,7 @@ def test_read_csv_categorical() -> None:
 def test_cat_to_dummies() -> None:
     df = pl.DataFrame({"foo": [1, 2, 3, 4], "bar": ["a", "b", "a", "c"]})
     df = df.with_columns(pl.col("bar").cast(pl.Categorical))
-    assert pl.get_dummies(df).to_dict(False) == {
+    assert df.to_dummies().to_dict(False) == {
         "foo_1": [1, 0, 0, 0],
         "foo_2": [0, 1, 0, 0],
         "foo_3": [0, 0, 1, 0],

py-polars/tests/unit/test_df.py

@@ -800,9 +800,9 @@ def test_arg_where() -> None:
     assert_series_equal(pl.arg_where(s, eager=True).cast(int), pl.Series([0, 2]))
 
 
-def test_get_dummies() -> None:
+def test_to_dummies2() -> None:
     df = pl.DataFrame({"a": [1, 2, 3]})
-    res = pl.get_dummies(df)
+    res = df.to_dummies()
     expected = pl.DataFrame(
         {"a_1": [1, 0, 0], "a_2": [0, 1, 0], "a_3": [0, 0, 1]}
     ).with_columns(pl.all().cast(pl.UInt8))
@@ -820,10 +820,16 @@ def test_get_dummies() -> None:
         },
         schema={"i": pl.Int32, "category|cat": pl.UInt8, "category|dog": pl.UInt8},
     )
-    result = pl.get_dummies(df, columns=["category"], separator="|")
+    result = df.to_dummies(columns=["category"], separator="|")
     assert_frame_equal(result, expected)
 
 
+def test_get_dummies_function_deprecated() -> None:
+    df = pl.DataFrame({"a": [1, 2, 3]})
+    with pytest.deprecated_call():
+        pl.get_dummies(df)
+
+
 def test_to_pandas(df: pl.DataFrame) -> None:
     # pyarrow cannot deal with unsigned dictionary integer yet.
     # pyarrow cannot convert a time64 w/ non-zero nanoseconds