Added python 2.7 support (untested)

pylearning/ensembles/ensembles.py

@@ -1,15 +1,16 @@
 import logging
-import random
 import abc
 import numpy as np
 from operator import itemgetter
 from concurrent.futures import ProcessPoolExecutor
 
+
 from ..trees import DecisionTreeRegressor
 from ..trees import DecisionTreeClassifier
+from six import add_metaclass
 
-
-class RandomForest(metaclass=abc.ABCMeta):
+@add_metaclass(abc.ABCMeta)
+class RandomForest(object):
     """
     Abstract base class for random forest algorithms. This class is not meant
     to be instanciated, ont its subclasses can be used.

pylearning/neighbours/neighbours.py

@@ -1,8 +1,9 @@
 import abc
 import numpy as np
+from six import add_metaclass
 
-
-class KNN(metaclass=abc.ABCMeta):
+@add_metaclass(abc.ABCMeta)
+class KNN(object):
     """
     Abstract base class for nearest neighbours algorithms. This class is not
     meant to be instanciated, ont its subclasses can be used.

pylearning/trees/trees.py

@@ -1,13 +1,14 @@
 import random
 import abc
 import numpy as np
-from math import log2, sqrt
+#from math import log2, sqrt
 from numbers import Number
 
 from .node import DecisionNode
+from six import add_metaclass
 
-
-class DecisionTree(metaclass=abc.ABCMeta):
+@add_metaclass(abc.ABCMeta)
+class DecisionTree(object):
     """
     Abstract base class for decision trees. This class is not meant to be
     instanciated,only its subclasses can be used.
@@ -68,9 +69,9 @@ def set_number_features_evaluated_split(self, row):
                 self.max_split_features <= len(row) else len(row)
         elif isinstance(self.max_split_features, str):
             if self.max_split_features in ['auto','sqrt']:
-                self.considered_features = int(sqrt(len(row)))
+                self.considered_features = int(np.sqrt(len(row)))
             elif self.max_split_features == 'log2':
-                self.considered_features = int(log2(len(row)))
+                self.considered_features = int(np.log2(len(row)))
         else:
             self.considered_features = len(row)
 
@@ -104,7 +105,7 @@ def entropy(self, targets):
         ent = 0.0
         for val in results.values():
             p = float(val) / len(targets)
-            ent -= p * log2(p)
+            ent -= p * np.log2(p)
         return ent
 
 

setup.cfg

@@ -0,0 +1,5 @@
+[bdist_wheel]
+universal = 1
+
+[metadata]
+license_file = LICENSE.md

setup.py

@@ -27,6 +27,7 @@
         'Intended Audience :: Developers',
         'Topic :: Scientific/Engineering :: Artificial Intelligence',
         'License :: OSI Approved :: MIT License',
+        'Programming Language :: Python :: 2.7',
         'Programming Language :: Python :: 3',
         'Programming Language :: Python :: 3.3',
         'Programming Language :: Python :: 3.4',
@@ -37,5 +38,6 @@
 
     packages=find_packages(exclude=['contrib','docs','tests']),
 
-    install_requires=['numpy']
+    install_requires=['numpy','six','futures;python_version<"3"']
+
 )