Updated util dependencies.

Author: Olcay Taner YILDIZ

Classification/InstanceList/InstanceList.py

@@ -58,6 +58,7 @@ class label. The function reads the file line by line and for each line; dependi
                     self.list = []
                     file = open(fileName, 'r', encoding='utf8')
                     lines = file.readlines()
+                    file.close()
                     for line in lines:
                         attributeList = line.strip().split(separator)
                         if len(attributeList) == listOrDefinition.attributeCount() + 1:

Classification/InstanceList/Partition.py

@@ -1,5 +1,7 @@
 import random
 
+from Util.RandomArray import RandomArray
+
 from Classification.InstanceList.InstanceList import InstanceList
 from Classification.InstanceList.InstanceListOfSameClass import InstanceListOfSameClass
 
@@ -35,9 +37,7 @@ def __init__(self, instanceList: InstanceList = None, ratio=None, seed=None, str
                     if stratified:
                         distribution = instanceList.classDistribution()
                         counts = [0] * len(distribution)
-                        randomArray = [i for i in range(instanceList.size())]
-                        random.seed(seed)
-                        random.shuffle(randomArray)
+                        randomArray = RandomArray.indexArray(instanceList.size(), seed)
                         for i in range(instanceList.size()):
                             instance = instanceList.get(randomArray[i])
                             classIndex = distribution.getIndex(instance.getClassLabel())

Classification/Model/DecisionTree/DecisionNode.py

@@ -1,4 +1,5 @@
 from Math.DiscreteDistribution import DiscreteDistribution
+from Util.RandomArray import RandomArray
 
 from Classification.Attribute.ContinuousAttribute import ContinuousAttribute
 from Classification.Attribute.DiscreteAttribute import DiscreteAttribute
@@ -67,12 +68,11 @@ def __init__(self, data: InstanceList, condition=None, parameter=None, isStump=F
             return
         if isStump and condition is not None:
             return
-        indexList = [i for i in range(data.get(0).attributeSize())]
         if parameter is not None and parameter.getAttributeSubsetSize() < data.get(0).attributeSize():
-            random.seed(parameter.getSeed())
-            random.shuffle(indexList)
+            indexList = RandomArray.indexArray(data.get(0).attributeSize(), parameter.getSeed())
             size = parameter.getAttributeSubsetSize()
         else:
+            indexList = [i for i in range(data.get(0).attributeSize())]
             size = data.get(0).attributeSize()
         classDistribution = data.classDistribution()
         bestEntropy = data.classDistribution().entropy()

setup.py

@@ -17,7 +17,7 @@
     author='olcaytaner',
     author_email='olcay.yildiz@ozyegin.edu.tr',
     description='Classification library',
-    install_requires=['NlpToolkit-Math', 'NlpToolkit-DataStructure', 'NlpToolkit-Sampling'],
+    install_requires=['NlpToolkit-Math', 'NlpToolkit-DataStructure', 'NlpToolkit-Sampling', 'NlpToolkit-Util'],
     long_description=long_description,
     long_description_content_type='text/markdown'
 )

test/Classifier/C45Test.py

@@ -20,10 +20,6 @@ def test_Train(self):
         self.assertAlmostEqual(6.48, 100 * c45.test(self.car.getInstanceList()).getErrorRate(), 2)
         c45.train(self.tictactoe.getInstanceList(), c45Parameter)
         self.assertAlmostEqual(8.77, 100 * c45.test(self.tictactoe.getInstanceList()).getErrorRate(), 2)
-        c45.train(self.nursery.getInstanceList(), c45Parameter)
-        self.assertAlmostEqual(2.31, 100 * c45.test(self.nursery.getInstanceList()).getErrorRate(), 2)
-        c45.train(self.chess.getInstanceList(), c45Parameter)
-        self.assertAlmostEqual(52.57, 100 * c45.test(self.chess.getInstanceList()).getErrorRate(), 2)
 
 
 if __name__ == '__main__':