DEPR: mad (pandas-dev#46707)

Author: rhshadrach

doc/source/whatsnew/v1.5.0.rst

@@ -430,7 +430,7 @@ Other Deprecations
 - Deprecated :attr:`Timedelta.freq` and :attr:`Timedelta.is_populated` (:issue:`46430`)
 - Deprecated :attr:`Timedelta.delta` (:issue:`46476`)
 - Deprecated the ``closed`` argument in :meth:`interval_range` in favor of ``inclusive`` argument; In a future version passing ``closed`` will raise (:issue:`40245`)
--
+- Deprecated the methods :meth:`DataFrame.mad`, :meth:`Series.mad`, and the corresponding groupby methods (:issue:`11787`)
 
 .. ---------------------------------------------------------------------------
 .. _whatsnew_150.performance:

pandas/core/generic.py

@@ -10881,6 +10881,9 @@ def mad(
         """
         {desc}
 
+        .. deprecated:: 1.5.0
+            mad is deprecated.
+
         Parameters
         ----------
         axis : {axis_descr}
@@ -10897,6 +10900,12 @@ def mad(
         {see_also}\
         {examples}
         """
+        msg = (
+            "The 'mad' method is deprecated and will be removed in a future version. "
+            "To compute the same result, you may do `(df - df.mean()).abs().mean()`."
+        )
+        warnings.warn(msg, FutureWarning, stacklevel=find_stack_level())
+
         if not is_bool(skipna):
             warnings.warn(
                 "Passing None for skipna is deprecated and will raise in a future"

pandas/tests/frame/test_reductions.py

@@ -67,6 +67,7 @@ def assert_stat_op_calc(
     skipna_alternative : function, default None
         NaN-safe version of alternative
     """
+    warn = FutureWarning if opname == "mad" else None
     f = getattr(frame, opname)
 
     if check_dates:
@@ -88,8 +89,9 @@ def wrapper(x):
             return alternative(x.values)
 
         skipna_wrapper = tm._make_skipna_wrapper(alternative, skipna_alternative)
-        result0 = f(axis=0, skipna=False)
-        result1 = f(axis=1, skipna=False)
+        with tm.assert_produces_warning(warn, match="The 'mad' method is deprecated"):
+            result0 = f(axis=0, skipna=False)
+            result1 = f(axis=1, skipna=False)
         tm.assert_series_equal(
             result0, frame.apply(wrapper), check_dtype=check_dtype, rtol=rtol, atol=atol
         )
@@ -102,8 +104,9 @@ def wrapper(x):
     else:
         skipna_wrapper = alternative
 
-    result0 = f(axis=0)
-    result1 = f(axis=1)
+    with tm.assert_produces_warning(warn, match="The 'mad' method is deprecated"):
+        result0 = f(axis=0)
+        result1 = f(axis=1)
     tm.assert_series_equal(
         result0,
         frame.apply(skipna_wrapper),
@@ -125,14 +128,18 @@ def wrapper(x):
         assert lcd_dtype == result1.dtype
 
     # bad axis
-    with pytest.raises(ValueError, match="No axis named 2"):
-        f(axis=2)
+    with tm.assert_produces_warning(warn, match="The 'mad' method is deprecated"):
+        with pytest.raises(ValueError, match="No axis named 2"):
+            f(axis=2)
 
     # all NA case
     if has_skipna:
         all_na = frame * np.NaN
-        r0 = getattr(all_na, opname)(axis=0)
-        r1 = getattr(all_na, opname)(axis=1)
+        with tm.assert_produces_warning(
+            warn, match="The 'mad' method is deprecated", raise_on_extra_warnings=False
+        ):
+            r0 = getattr(all_na, opname)(axis=0)
+            r1 = getattr(all_na, opname)(axis=1)
         if opname in ["sum", "prod"]:
             unit = 1 if opname == "prod" else 0  # result for empty sum/prod
             expected = Series(unit, index=r0.index, dtype=r0.dtype)
@@ -167,9 +174,13 @@ class TestDataFrameAnalytics:
         ],
     )
     def test_stat_op_api_float_string_frame(self, float_string_frame, axis, opname):
-        getattr(float_string_frame, opname)(axis=axis)
-        if opname not in ("nunique", "mad"):
-            getattr(float_string_frame, opname)(axis=axis, numeric_only=True)
+        warn = FutureWarning if opname == "mad" else None
+        with tm.assert_produces_warning(
+            warn, match="The 'mad' method is deprecated", raise_on_extra_warnings=False
+        ):
+            getattr(float_string_frame, opname)(axis=axis)
+            if opname not in ("nunique", "mad"):
+                getattr(float_string_frame, opname)(axis=axis, numeric_only=True)
 
     @pytest.mark.filterwarnings("ignore:Dropping of nuisance:FutureWarning")
     @pytest.mark.parametrize("axis", [0, 1])
@@ -1424,7 +1435,9 @@ def test_frame_any_with_timedelta(self):
     def test_reductions_deprecation_skipna_none(self, frame_or_series):
         # GH#44580
         obj = frame_or_series([1, 2, 3])
-        with tm.assert_produces_warning(FutureWarning, match="skipna"):
+        with tm.assert_produces_warning(
+            FutureWarning, match="skipna", raise_on_extra_warnings=False
+        ):
             obj.mad(skipna=None)
 
     def test_reductions_deprecation_level_argument(
@@ -1445,7 +1458,7 @@ def test_reductions_skipna_none_raises(
                 pytest.mark.xfail(reason="Count does not accept skipna")
             )
         elif reduction_functions == "mad":
-            pytest.skip("Mad needs a deprecation cycle: GH 11787")
+            pytest.skip("Mad is deprecated: GH#11787")
         obj = frame_or_series([1, 2, 3])
         msg = 'For argument "skipna" expected type bool, received type NoneType.'
         with pytest.raises(ValueError, match=msg):
@@ -1644,25 +1657,37 @@ def test_mad_nullable_integer(any_signed_int_ea_dtype):
     df = DataFrame(np.random.randn(100, 4).astype(np.int64))
     df2 = df.astype(any_signed_int_ea_dtype)
 
-    result = df2.mad()
-    expected = df.mad()
+    with tm.assert_produces_warning(
+        FutureWarning, match="The 'mad' method is deprecated"
+    ):
+        result = df2.mad()
+        expected = df.mad()
     tm.assert_series_equal(result, expected)
 
-    result = df2.mad(axis=1)
-    expected = df.mad(axis=1)
+    with tm.assert_produces_warning(
+        FutureWarning, match="The 'mad' method is deprecated"
+    ):
+        result = df2.mad(axis=1)
+        expected = df.mad(axis=1)
     tm.assert_series_equal(result, expected)
 
     # case with NAs present
     df2.iloc[::2, 1] = pd.NA
 
-    result = df2.mad()
-    expected = df.mad()
-    expected[1] = df.iloc[1::2, 1].mad()
+    with tm.assert_produces_warning(
+        FutureWarning, match="The 'mad' method is deprecated"
+    ):
+        result = df2.mad()
+        expected = df.mad()
+        expected[1] = df.iloc[1::2, 1].mad()
     tm.assert_series_equal(result, expected)
 
-    result = df2.mad(axis=1)
-    expected = df.mad(axis=1)
-    expected[::2] = df.T.loc[[0, 2, 3], ::2].mad()
+    with tm.assert_produces_warning(
+        FutureWarning, match="The 'mad' method is deprecated"
+    ):
+        result = df2.mad(axis=1)
+        expected = df.mad(axis=1)
+        expected[::2] = df.T.loc[[0, 2, 3], ::2].mad()
     tm.assert_series_equal(result, expected)
 
 
@@ -1675,8 +1700,11 @@ def test_mad_nullable_integer_all_na(any_signed_int_ea_dtype):
     # case with all-NA row/column
     df2.iloc[:, 1] = pd.NA  # FIXME(GH#44199): this doesn't operate in-place
     df2.iloc[:, 1] = pd.array([pd.NA] * len(df2), dtype=any_signed_int_ea_dtype)
-    result = df2.mad()
-    expected = df.mad()
+    with tm.assert_produces_warning(
+        FutureWarning, match="The 'mad' method is deprecated"
+    ):
+        result = df2.mad()
+        expected = df.mad()
     expected[1] = pd.NA
     expected = expected.astype("Float64")
     tm.assert_series_equal(result, expected)

pandas/tests/groupby/aggregate/test_aggregate.py

@@ -1325,11 +1325,13 @@ def test_groupby_aggregate_directory(reduction_func):
     # GH#32793
     if reduction_func in ["corrwith", "nth"]:
         return None
+    warn = FutureWarning if reduction_func == "mad" else None
 
     obj = DataFrame([[0, 1], [0, np.nan]])
 
-    result_reduced_series = obj.groupby(0).agg(reduction_func)
-    result_reduced_frame = obj.groupby(0).agg({1: reduction_func})
+    with tm.assert_produces_warning(warn, match="The 'mad' method is deprecated"):
+        result_reduced_series = obj.groupby(0).agg(reduction_func)
+        result_reduced_frame = obj.groupby(0).agg({1: reduction_func})
 
     if reduction_func in ["size", "ngroup"]:
         # names are different: None / 1

pandas/tests/groupby/test_allowlist.py

@@ -178,6 +178,7 @@ def test_regression_allowlist_methods(raw_frame, op, level, axis, skipna, sort):
     # GH6944
     # GH 17537
     # explicitly test the allowlist methods
+    warn = FutureWarning if op == "mad" else None
 
     if axis == 0:
         frame = raw_frame
@@ -186,7 +187,8 @@ def test_regression_allowlist_methods(raw_frame, op, level, axis, skipna, sort):
 
     if op in AGG_FUNCTIONS_WITH_SKIPNA:
         grouped = frame.groupby(level=level, axis=axis, sort=sort)
-        result = getattr(grouped, op)(skipna=skipna)
+        with tm.assert_produces_warning(warn, match="The 'mad' method is deprecated"):
+            result = getattr(grouped, op)(skipna=skipna)
         with tm.assert_produces_warning(FutureWarning):
             expected = getattr(frame, op)(level=level, axis=axis, skipna=skipna)
         if sort:

pandas/tests/groupby/test_apply.py

@@ -1047,6 +1047,8 @@ def test_apply_with_timezones_aware():
 def test_apply_is_unchanged_when_other_methods_are_called_first(reduction_func):
     # GH #34656
     # GH #34271
+    warn = FutureWarning if reduction_func == "mad" else None
+
     df = DataFrame(
         {
             "a": [99, 99, 99, 88, 88, 88],
@@ -1068,7 +1070,8 @@ def test_apply_is_unchanged_when_other_methods_are_called_first(reduction_func):
     # Check output when another method is called before .apply()
     grp = df.groupby(by="a")
     args = {"nth": [0], "corrwith": [df]}.get(reduction_func, [])
-    _ = getattr(grp, reduction_func)(*args)
+    with tm.assert_produces_warning(warn, match="The 'mad' method is deprecated"):
+        _ = getattr(grp, reduction_func)(*args)
     result = grp.apply(sum)
     tm.assert_frame_equal(result, expected)
 

pandas/tests/groupby/test_categorical.py

@@ -1357,6 +1357,7 @@ def test_series_groupby_on_2_categoricals_unobserved(reduction_func, observed, r
             reason="TODO: implemented SeriesGroupBy.corrwith. See GH 32293"
         )
         request.node.add_marker(mark)
+    warn = FutureWarning if reduction_func == "mad" else None
 
     df = DataFrame(
         {
@@ -1371,7 +1372,8 @@ def test_series_groupby_on_2_categoricals_unobserved(reduction_func, observed, r
 
     series_groupby = df.groupby(["cat_1", "cat_2"], observed=observed)["value"]
     agg = getattr(series_groupby, reduction_func)
-    result = agg(*args)
+    with tm.assert_produces_warning(warn, match="The 'mad' method is deprecated"):
+        result = agg(*args)
 
     assert len(result) == expected_length
 
@@ -1390,6 +1392,7 @@ def test_series_groupby_on_2_categoricals_unobserved_zeroes_or_nans(
             reason="TODO: implemented SeriesGroupBy.corrwith. See GH 32293"
         )
         request.node.add_marker(mark)
+    warn = FutureWarning if reduction_func == "mad" else None
 
     df = DataFrame(
         {
@@ -1403,7 +1406,8 @@ def test_series_groupby_on_2_categoricals_unobserved_zeroes_or_nans(
 
     series_groupby = df.groupby(["cat_1", "cat_2"], observed=False)["value"]
     agg = getattr(series_groupby, reduction_func)
-    result = agg(*args)
+    with tm.assert_produces_warning(warn, match="The 'mad' method is deprecated"):
+        result = agg(*args)
 
     zero_or_nan = _results_for_groupbys_with_missing_categories[reduction_func]
 
@@ -1426,6 +1430,7 @@ def test_dataframe_groupby_on_2_categoricals_when_observed_is_true(reduction_fun
     # does not return the categories that are not in df when observed=True
     if reduction_func == "ngroup":
         pytest.skip("ngroup does not return the Categories on the index")
+    warn = FutureWarning if reduction_func == "mad" else None
 
     df = DataFrame(
         {
@@ -1439,7 +1444,8 @@ def test_dataframe_groupby_on_2_categoricals_when_observed_is_true(reduction_fun
     df_grp = df.groupby(["cat_1", "cat_2"], observed=True)
 
     args = {"nth": [0], "corrwith": [df]}.get(reduction_func, [])
-    res = getattr(df_grp, reduction_func)(*args)
+    with tm.assert_produces_warning(warn, match="The 'mad' method is deprecated"):
+        res = getattr(df_grp, reduction_func)(*args)
 
     for cat in unobserved_cats:
         assert cat not in res.index
@@ -1456,6 +1462,7 @@ def test_dataframe_groupby_on_2_categoricals_when_observed_is_false(
 
     if reduction_func == "ngroup":
         pytest.skip("ngroup does not return the Categories on the index")
+    warn = FutureWarning if reduction_func == "mad" else None
 
     df = DataFrame(
         {
@@ -1469,7 +1476,8 @@ def test_dataframe_groupby_on_2_categoricals_when_observed_is_false(
     df_grp = df.groupby(["cat_1", "cat_2"], observed=observed)
 
     args = {"nth": [0], "corrwith": [df]}.get(reduction_func, [])
-    res = getattr(df_grp, reduction_func)(*args)
+    with tm.assert_produces_warning(warn, match="The 'mad' method is deprecated"):
+        res = getattr(df_grp, reduction_func)(*args)
 
     expected = _results_for_groupbys_with_missing_categories[reduction_func]
 

pandas/tests/groupby/test_function.py

@@ -321,11 +321,17 @@ def test_mad(self, gb, gni):
         # mad
         expected = DataFrame([[0], [np.nan]], columns=["B"], index=[1, 3])
         expected.index.name = "A"
-        result = gb.mad()
+        with tm.assert_produces_warning(
+            FutureWarning, match="The 'mad' method is deprecated"
+        ):
+            result = gb.mad()
         tm.assert_frame_equal(result, expected)
 
         expected = DataFrame([[1, 0.0], [3, np.nan]], columns=["A", "B"], index=[0, 1])
-        result = gni.mad()
+        with tm.assert_produces_warning(
+            FutureWarning, match="The 'mad' method is deprecated"
+        ):
+            result = gni.mad()
         tm.assert_frame_equal(result, expected)
 
     def test_describe(self, df, gb, gni):