统计学习方法 第五章 课后习题

5.1 根据表5.1所给的训练数据集，利用信息增益比（C4.5算法）生成决策树

DTree.py实现了ID3、C4.5做树，而CART只是实现计算Gini肯尼指数，没实现做树，步骤是一样的，我进度拖得比较慢，就没有实现，基本上可以套用。你们也可直接计算下，我计算过一遍然后在直接写代码

# -*- coding: utf-8 -*-
import math
C45_Flag = True#算法标志
ID3_Flag = False

class DtreeStruct:
    def __init__(self,next_nodelist=None,Ai=None,Aivalue=None,ck=None,value=None):
        self.next_nodelist = next_nodelist
        self.Ai = Ai
        self.Aival= Aivalue
        self.value = value
        self.ck = ck
        
    def Print(self):
        def Fprint(node):
            if node.next_nodelist == None:
                print("叶节点：",node.Ai,node.Aival,node.value,node.ck)
                return
            print("节点",node.next_nodelist,node.Ai,node.Aival,node.value,node.ck)
            for node in node.next_nodelist:
                Fprint(node) 
                
        print("根节点",self.next_nodelist,self.Ai,self.Aival,self.value,self.ck)
        for node in self.next_nodelist:
            Fprint(node)
            
class DTree: 
    def __init__(self,datasets,labels):
        self.tree = None
        self.datasets = datasets
        self.labels = labels
        self.GetAandC()
        
        
    def GetAandC(self):
        self.C = {}
        self.A = {}
        self.C[self.labels[-1]] = set([ line[-1] for line in self.datasets])
        for i in range(len(self.labels)-1):
            self.A[self.labels[i]] = set([ line[i] for line in self.datasets])
    
    def ID3CreateTree(self,epsilon,ICflag):
        #经验熵
        def emp_entropy(data,label):
            dic = {}
            datalen = len(data)
            indx = self.labels.index(label)
            for line in data:
                if line[indx] not in dic:
                    dic[line[indx]] = 0
                #该特征下的取值分类个数（基本上为‘类别’）
                dic[line[indx]] += 1
            return -sum([(dic[p]/datalen)*math.log(dic[p]/datalen,2) for p in dic.keys()])
        
        #经验条件熵
        def emp_cdtl_entropy(data,Ai):
            dic = {}
            data_dic = {}
            c = list(self.C.keys())[0]
            indx = self.labels.index(Ai)
            datalen = len(data)
            for line in data: