PERF: fix some of .clip() performance regression by using numpy arrays where possible (#24735)

Author: qwhelan

asv_bench/benchmarks/series_methods.py

@@ -140,11 +140,13 @@ def time_map(self, mapper):
 
 
 class Clip(object):
+    params = [50, 1000, 10**5]
+    param_names = ['n']
 
-    def setup(self):
-        self.s = Series(np.random.randn(50))
+    def setup(self, n):
+        self.s = Series(np.random.randn(n))
 
-    def time_clip(self):
+    def time_clip(self, n):
         self.s.clip(0, 1)
 
 

pandas/core/generic.py

@@ -7148,12 +7148,18 @@ def _clip_with_scalar(self, lower, upper, inplace=False):
             raise ValueError("Cannot use an NA value as a clip threshold")
 
         result = self
-        if upper is not None:
-            subset = self.le(upper, axis=None) | isna(result)
-            result = result.where(subset, upper, axis=None, inplace=False)
-        if lower is not None:
-            subset = self.ge(lower, axis=None) | isna(result)
-            result = result.where(subset, lower, axis=None, inplace=False)
+        mask = isna(self.values)
+
+        with np.errstate(all='ignore'):
+            if upper is not None:
+                subset = self.to_numpy() <= upper
+                result = result.where(subset, upper, axis=None, inplace=False)
+            if lower is not None:
+                subset = self.to_numpy() >= lower
+                result = result.where(subset, lower, axis=None, inplace=False)
+
+        if np.any(mask):
+            result[mask] = np.nan
 
         if inplace:
             self._update_inplace(result)