Numpy always faster in real-world use

choicemodels/tools/mergedchoicetable.py

@@ -325,21 +325,12 @@ def _build_table(self):
         # replacement, with optional alternative-specific weights but NOT weights that 
         # apply to combinations of observation x alternative
 
-        # No weights
         if (self.replace == True) & (self.weights is None):
 
-            if (sys.version_info[0] >= 3) & (sys.version_info[1] >= 6):
-                # 'random.choices' was added in python 3.6, and in my testing it's about
-                # 2x faster than 'np.random.choice' for unweighted sampling without
-                # replacement, as of numpy 1.13
-                alt_ids = random.choices(self.alternatives.index.values, 
-                                         k = n_obs * samp_size)
-            else:
-                alt_ids = np.random.choice(self.alternatives.index.values, 
-                                           replace = True,
-                                           size = n_obs * samp_size)
+            alt_ids = np.random.choice(self.alternatives.index.values, 
+                                       replace = True,
+                                       size = n_obs * samp_size)
 
-        # Alternative-specific weights: numpy is most efficient
         elif (self.replace == True) & (self.weights_1d == True):
 
             alt_ids = np.random.choice(self.alternatives.index.values,