Parametrize asv; temp. if-cond to test against master

asv_bench/benchmarks/frame_methods.py

@@ -412,8 +412,10 @@ def time_frame_nunique(self):
 class Duplicated(object):
 
     goal_time = 0.2
+    params = (['first', 'last'], [True, False])
+    param_names = ['keep', 'return_inverse']
 
-    def setup(self):
+    def setup(self, keep, return_inverse):
         n = (1 << 20)
         t = date_range('2015-01-01', freq='S', periods=(n // 64))
         xs = np.random.randn(n // 64).round(2)
@@ -428,24 +430,23 @@ def setup(self):
         df3.loc[:, 'F'] = Series('', index=df3.index).str.cat(df3.astype(str))
         self.df3 = df3
 
-    def time_frame_duplicated(self):
-        self.df.duplicated()
-
-    def time_frame_duplicated_wide(self):
-        self.df2.duplicated()
-
-    def time_frame_duplicated_wide_inverse(self):
-        # will take fastpath for no duplicates
-        self.df2.duplicated(return_inverse=True)
-
-    def time_frame_duplicated_mixed(self):
-        self.df3.duplicated()
-
-    def time_frame_duplicated_mixed_inverse(self):
-        self.df3.duplicated(return_inverse=True)
-
-    def time_frame_duplicated_mixed_inverse_last(self):
-        self.df3.duplicated(return_inverse=True, keep='last')
+    def time_frame_duplicated(self, keep, return_inverse):
+        if return_inverse:
+            self.df.duplicated(keep=keep, return_inverse=return_inverse)
+        else:
+            self.df.duplicated(keep=keep)
+
+    def time_frame_duplicated_wide(self, keep, return_inverse):
+        if return_inverse:
+            self.df2.duplicated(keep=keep, return_inverse=return_inverse)
+        else:
+            self.df2.duplicated(keep=keep)
+
+    def time_frame_duplicated_mixed(self, keep, return_inverse):
+        if return_inverse:
+            self.df3.duplicated(keep=keep, return_inverse=return_inverse)
+        else:
+            self.df3.duplicated(keep=keep)
 
 
 class XS(object):