fix failing metrics tests under python3.5

Author: lucaskolstad

.gitignore

@@ -2,3 +2,4 @@
 build/
 dist/
 mla.egg-info/
+.cache

mla/metrics/base.py

@@ -4,10 +4,10 @@
 
 
 def check_data(a, b):
-    if isinstance(a, list):
+    if isinstance(a, list) or isinstance(a, range):
         a = np.array(a)
 
-    if isinstance(b, list):
+    if isinstance(b, list) or isinstance(b, range):
         b = np.array(b)
 
     if type(a) != type(b):

mla/metrics/metrics.py

@@ -60,7 +60,6 @@ def root_mean_squared_log_error(actual, predicted):
 
 def logloss(actual, predicted):
     predicted = np.clip(predicted, EPS, 1 - EPS)
-    predicted /= predicted.sum(axis=1, keepdims=True)
     loss = -np.sum(actual * np.log(predicted))
     return loss / float(actual.shape[0])