def crossValidation(trainSet, testSet, labels, m):
"""
:param trainSet:训练集
:param testSet:测试集
:param labels:特征类别属性
:param m:先验概率对后验概率的影响因子
:return:该模型的性能指标
"""
v = len(trainSet)
TreeSet = []
for i in range(v): # 分别进行K次生成决策树、决策树剪枝的过程
temData = tuple(trainSet[i])
Tree = createC4_5Tree(list(temData), list(labels))
purnTree = PostPruning_IMEP(Tree, labels, trainSet[i], m)
# print('purnTree=', purnTree)
TreeSet.append(purnTree)
errorSum = 0.0
for i in range(0, v): # 将剪枝后的决策树进行错误估计
error = leafError(TreeSet[i], labels, testSet[i])
errorSum += error
return errorSum / v
def showDT(dataSet, labels):
"""
:param dataSet:数据集
:param labels:属性标签
"""
# ID3算法生成分类决策树
ID3Tree = createID3Tree(list(dataSet), list(labels))
print('The ID3 Decision Tree is', ID3Tree)
# C4.5算法生成分类决策树
C4_5Tree = createC4_5Tree(list(dataSet), list(labels))
print('The C4.5 Decision Tree is', C4_5Tree)
# CART算法生成分类决策树
CARTTree = createCARTTree(list(dataSet), list(labels))
print('The CART Decision Tree is', CARTTree)
# 显示各个决策树
createPlot(ID3Tree, 'ID3 Decision Tree')
createPlot(C4_5Tree, 'C4.5 Decision Tree')
createPlot(CARTTree, 'CART Decision Tree')
plt.show() # 显示决策树
# 计算剪枝前后准确度
ID3Accuracy = testAccuracy(ID3TrainTree, labels, trainSet)
print('The Accuracy of ID3 Decision Tree is', ID3Accuracy)
PreID3Accuracy = testAccuracy(PreID3Tree, labels, trainSet)
print('The Accuracy of Pre_Pruning_ID3 Decision Tree is', PreID3Accuracy)
REPID3Accuracy = testAccuracy(REPID3Tree, labels, trainSet)
print('The Accuracy of REP_Pruning_ID3 Decision Tree is', REPID3Accuracy)
PEPID3Accuracy = testAccuracy(PEPID3Tree, labels, trainSet)
print('The Accuracy of PEP_Pruning_ID3 Decision Tree is', PEPID3Accuracy)
"""对C4.5算法生成的决策树，进行剪枝"""
''''''
print('===========================================')
# 先用C4.5算法，生成决策树
C4_5Tree = createC4_5Tree(list(trainSet), list(labels))
print('The C4.5 Decision Tree:', 'Depth:', getTreeDepth(C4_5Tree), ';Leaf:', getNumLeaf(C4_5Tree))
print('The Node with largest number is', findKeyNode(C4_5Tree))
copyC4_5Tree = copy.deepcopy(C4_5Tree) # 将生成树拷贝
copyC4_5Tree1 = copy.deepcopy(C4_5Tree) # 将生成树拷贝
# 使用PEP方法，对决策树进行后剪枝
PEPC4_5Tree = PostPruning_PEP(C4_5Tree, list(labels), list(trainSet))
print('The PEP_Pruning_C4.5 Decision Tree:', 'Depth:', getTreeDepth(PEPC4_5Tree), ';Leaf:', getNumLeaf(PEPC4_5Tree))
print('The Node with largest number is', findKeyNode(PEPC4_5Tree))
# 使用MEP方法，对决策树进行后剪枝
MEPC4_5Tree = PostPruning_MEP(copyC4_5Tree, list(labels), list(trainSet), k)
print('The MEP_Pruning_C4.5 Decision Tree:', 'Depth:', getTreeDepth(MEPC4_5Tree), ';Leaf:', getNumLeaf(MEPC4_5Tree))
print('The Node with largest number is', findKeyNode(MEPC4_5Tree))