pct change bug issue 21200

pandas/core/groupby/groupby.py

@@ -2075,12 +2075,19 @@ def pct_change(self, periods=1, fill_method='pad', limit=None, freq=None,
                                                      fill_method=fill_method,
                                                      limit=limit, freq=freq,
                                                      axis=axis))
+        if fill_method:
+            new = DataFrameGroupBy(self._obj_with_exclusions,
+                                   grouper=self.grouper)
+            new.obj = getattr(new, fill_method)(limit=limit)
+            new._reset_cache()
+        else:
+            new = self
 
-        filled = getattr(self, fill_method)(limit=limit).drop(
-            self.grouper.names, axis=1)
-        shifted = filled.shift(periods=periods, freq=freq)
+        obj = new.obj.drop(self.grouper.names, axis=1)
+        shifted = new.shift(periods=periods, freq=freq).\
+            drop(self.grouper.names, axis=1)
+        return (obj / shifted) - 1
 
-        return (filled / shifted) - 1
 
     @Substitution(name='groupby')
     @Appender(_doc_template)
@@ -3942,11 +3949,16 @@ def _apply_to_column_groupbys(self, func):
         return func(self)
 
     def pct_change(self, periods=1, fill_method='pad', limit=None, freq=None):
-        """Calculate percent change of each value to previous entry in group"""
-        filled = getattr(self, fill_method)(limit=limit)
-        shifted = filled.shift(periods=periods, freq=freq)
+        """Calcuate pct_change of each value to previous entry in group"""
+        if fill_method:
+            new = SeriesGroupBy(self.obj, grouper=self.grouper)
+            new.obj = getattr(new, fill_method)(limit=limit)
+            new._reset_cache()
+        else:
+            new = self
 
-        return (filled / shifted) - 1
+        shifted = new.shift(periods=periods, freq=freq)
+        return (new.obj / shifted) - 1
 
 
 class NDFrameGroupBy(GroupBy):

pandas/tests/groupby/test_transform.py

@@ -722,19 +722,30 @@ def interweave(list_obj):
 
 
 @pytest.mark.parametrize("test_series", [True, False])
+@pytest.mark.parametrize("shuffle", [True, False])
 @pytest.mark.parametrize("periods,fill_method,limit", [
     (1, 'ffill', None), (1, 'ffill', 1),
     (1, 'bfill', None), (1, 'bfill', 1),
     (-1, 'ffill', None), (-1, 'ffill', 1),
-    (-1, 'bfill', None), (-1, 'bfill', 1)])
-def test_pct_change(test_series, periods, fill_method, limit):
-    vals = [np.nan, np.nan, 1, 2, 4, 10, np.nan, np.nan]
-    exp_vals = Series(vals).pct_change(periods=periods,
-                                       fill_method=fill_method,
-                                       limit=limit).tolist()
-
-    df = DataFrame({'key': ['a'] * len(vals) + ['b'] * len(vals),
-                    'vals': vals * 2})
+    (-1, 'bfill', None), (-1, 'bfill', 1),
+])
+def test_pct_change(test_series, shuffle, periods, fill_method, limit):
+    vals = [3, np.nan, 1, 2, 4, 10, np.nan, 4]
+    keys = ['a', 'b']
+    key_v = np.repeat(keys, len(vals))
+    df = DataFrame({'key': key_v, 'vals': vals * 2})
+    if shuffle:
+        order = np.random.RandomState(seed=42).permutation(len(df))
+        df = df.reindex(order).reset_index(drop=True)
+
+    manual_apply = []
+    for k in keys:
+        ind = df.loc[df.key == k, 'vals']
+        manual_apply.append(ind.pct_change(periods=periods,
+                                           fill_method=fill_method,
+                                           limit=limit))
+    exp_vals = pd.concat(manual_apply, ignore_index=True)
+    exp = pd.DataFrame(exp_vals.values, columns=['A'])
     grp = df.groupby('key')
 
     def get_result(grp_obj):
@@ -743,14 +754,21 @@ def get_result(grp_obj):
                                   limit=limit)
 
     if test_series:
-        exp = pd.Series(exp_vals * 2)
-        exp.name = 'vals'
+        exp = exp.loc[:, 'A']
         grp = grp['vals']
         result = get_result(grp)
+        df.insert(0, 'A', result)
+        result = df.sort_values(by='key')
+        result = result.loc[:, 'A']
+        result = result.reset_index(drop=True)
         tm.assert_series_equal(result, exp)
     else:
-        exp = DataFrame({'vals': exp_vals * 2})
         result = get_result(grp)
+        result = result.reset_index(drop=True)
+        df.insert(0, 'A', result)
+        result = df.sort_values(by='key')
+        result = result.loc[:, ['A']]
+        result = result.reset_index(drop=True)
         tm.assert_frame_equal(result, exp)