PERF: DataFrame.clip / Series.clip

asv_bench/benchmarks/frame_methods.py

@@ -17,6 +17,24 @@
 from .pandas_vb_common import tm
 
 
+class Clip:
+    params = [
+        ["float64", "Float64", "float64[pyarrow]"],
+        [True, False],
+    ]
+    param_names = ["dtype", "hasna"]
+
+    def setup(self, dtype, hasna):
+        data = np.random.randn(100_000, 10)
+        df = DataFrame(data, dtype=dtype)
+        if hasna:
+            df.iloc[2, :] = None
+        self.df = df
+
+    def time_clip(self, dtype, hasna):
+        self.df.clip(-1.0, 1.0)
+
+
 class GetNumericData:
     def setup(self):
         self.df = DataFrame(np.random.randn(10000, 25))

doc/source/whatsnew/v2.0.0.rst

@@ -1115,6 +1115,7 @@ Performance improvements
 - Performance improvement in :func:`merge` and :meth:`DataFrame.join` when joining on a sorted :class:`MultiIndex` (:issue:`48504`)
 - Performance improvement in :func:`to_datetime` when parsing strings with timezone offsets (:issue:`50107`)
 - Performance improvement in :meth:`DataFrame.loc` and :meth:`Series.loc` for tuple-based indexing of a :class:`MultiIndex` (:issue:`48384`)
+- Performance improvement in :meth:`DataFrame.clip` and :meth:`Series.clip` (:issue:`51472`)
 - Performance improvement for :meth:`Series.replace` with categorical dtype (:issue:`49404`)
 - Performance improvement for :meth:`MultiIndex.unique` (:issue:`48335`)
 - Performance improvement for indexing operations with nullable and arrow dtypes (:issue:`49420`, :issue:`51316`)

pandas/core/generic.py

@@ -7989,20 +7989,16 @@ def _clip_with_scalar(self, lower, upper, inplace: bool_t = False):
         ):
             raise ValueError("Cannot use an NA value as a clip threshold")
 
-        result = self
-        mask = isna(self._values)
-
-        with np.errstate(all="ignore"):
+        def blk_func(arr, lower=lower, upper=upper):
             if upper is not None:
-                subset = self <= upper
-                result = result.where(subset, upper, axis=None, inplace=False)
+                arr[arr > upper] = upper
             if lower is not None:
-                subset = self >= lower
-                result = result.where(subset, lower, axis=None, inplace=False)
-
-        if np.any(mask):
-            result[mask] = np.nan
+                arr[arr < lower] = lower
+            return arr
 
+        result = self if inplace else self.copy()
+        result_mgr = result._mgr.apply(blk_func)
+        result = self._constructor(result_mgr).__finalize__(self)
         if inplace:
             return self._update_inplace(result)
         else:

pandas/tests/copy_view/test_clip.py

@@ -12,10 +12,10 @@ def test_clip_inplace_reference(using_copy_on_write):
     view = df[:]
     df.clip(lower=2, inplace=True)
 
-    # Clip not actually inplace right now but could be
-    assert not np.shares_memory(get_array(df, "a"), arr_a)
-
     if using_copy_on_write:
+        assert not np.shares_memory(get_array(df, "a"), arr_a)
         assert df._mgr._has_no_reference(0)
         assert view._mgr._has_no_reference(0)
         tm.assert_frame_equal(df_copy, view)
+    else:
+        assert np.shares_memory(get_array(df, "a"), arr_a)