Source Code for Module rdkit.ML.DecTree.DecTree

  1  # 
  2  #  Copyright (C) 2000-2004  greg Landrum and Rational Discovery LLC 
  3  #  All Rights Reserved 
  4  # 
  5  """ Defines the class _DecTreeNode_, used to represent decision trees 
  6   
  7    _DecTreeNode_ is derived from _Tree.TreeNode_ 
  8   
  9  """ 
 10  from rdkit.ML.DecTree import Tree 
 11   
 12 -class DecTreeNode(Tree.TreeNode): 
 13    """ This is used to represent decision trees 
 14   
 15     _DecTreeNode_s are simultaneously the roots and branches of decision trees. 
 16     Everything is nice and recursive. 
 17   
 18     _DecTreeNode_s can save the following pieces of internal state, accessible via 
 19       standard setter/getter functions: 
 20   
 21       1) _Examples_: a list of examples which have been classified 
 22   
 23       2) _BadExamples_: a list of examples which have been misclassified 
 24   
 25       3) _TrainingExamples_: the list of examples used to train the tree 
 26   
 27       4) _TestExamples_: the list of examples used to test the tree 
 28      
 29    """ 
 30 -  def __init__(self,*args,**kwargs): 
 31      apply(Tree.TreeNode.__init__,(self,)+args,kwargs) 
 32      self.examples = [] 
 33      self.badExamples = [] 
 34      self.trainingExamples = [] 
 35      self.testExamples = [] 
 36 -  def ClassifyExample(self,example,appendExamples=0): 
 37      """ Recursively classify an example by running it through the tree 
 38       
 39        **Arguments** 
 40   
 41          - example: the example to be classified 
 42   
 43          - appendExamples: if this is nonzero then this node (and all children) 
 44            will store the example 
 45   
 46        **Returns** 
 47   
 48          the classification of _example_ 
 49   
 50        **NOTE:** 
 51          In the interest of speed, I don't use accessor functions 
 52          here.  So if you subclass DecTreeNode for your own trees, you'll 
 53          have to either include ClassifyExample or avoid changing the names 
 54          of the instance variables this needs. 
 55           
 56      """ 
 57      if appendExamples: 
 58        self.examples.append(example) 
 59      if self.terminalNode: 
 60        return self.label 
 61      else: 
 62        val = example[self.label] 
 63        return self.children[val].ClassifyExample(example,appendExamples) 
 64    
 65 -  def AddChild(self,name,label=None,data=None,isTerminal=0): 
 66      """ Constructs and adds a child with the specified data to our list 
 67   
 68        **Arguments** 
 69   
 70         - name: the name of the new node 
 71   
 72         - label: the label of the new node (should be an integer) 
 73   
 74         - data: the data to be stored in the new node 
 75   
 76         - isTerminal: a toggle to indicate whether or not the new node is 
 77           a terminal (leaf) node. 
 78   
 79        **Returns* 
 80   
 81          the _DecTreeNode_ which is constructed 
 82           
 83      """ 
 84      child = DecTreeNode(self,name,label,data,level=self.level+1,isTerminal=isTerminal) 
 85      self.children.append(child) 
 86      return child 
 87   
 88 -  def GetExamples(self): 
 89      return self.examples 
 90 -  def SetExamples(self,examples): 
 91      self.examples = examples 
 92   
 93 -  def GetBadExamples(self): 
 94      return self.badExamples 
 95 -  def SetBadExamples(self,examples): 
 96      self.badExamples = examples 
 97   
 98 -  def GetTrainingExamples(self): 
 99      return self.trainingExamples 
100 -  def SetTrainingExamples(self,examples): 
101      self.trainingExamples = examples 
102   
103 -  def GetTestExamples(self): 
104      return self.testExamples 
105 -  def SetTestExamples(self,examples): 
106      self.testExamples = examples 
107   
108 -  def ClearExamples(self): 
109      self.examples = [] 
110      self.badExamples = [] 
111      self.trainingExamples = [] 
112      self.testExamples = [] 
113      for child in self.GetChildren(): 
114        child.ClearExamples() 
115