Update logistic_regression.py (#64)

Correction to the cost function: since the function is J(\theta) = 1/m \sum -ylog(h) - (1-y)*(1-h), the cost variable should be set to cost_1 - cost_2, not cost_1 + cost_2

code/logistic_regression.py

@@ -33,7 +33,7 @@ def cost_function(features, labels, weights):
     class2_cost = (1-labels)*np.log(1-predictions)
 
     #Take the sum of both costs
-    cost = class1_cost + class2_cost
+    cost = class1_cost - class2_cost
 
     #Take the average cost
     cost = cost.sum() / observations