def demo():
print '... demo'
myDat, featNames = createDataSet()
print myDat
shannonEnt = decisionTree.calcShannonEnt(myDat)
print '当前数据集的熵是: ', shannonEnt
print '... 测试拆分数据集'
print decisionTree.splitDataSet(myDat, 0, 1)
print '... 测试最佳特征选择'
bestFeature = decisionTree.chooseBestFeatureToSplit(myDat)
print '最好的分类特征是 %s' % bestFeature
print '... 测试决策树的生成'
myTree = decisionTree.createTree(myDat, featNames)
print '生成的决策树是: \n', myTree
print '... 测试SortedCount'
classList = ['a', 'b', 'b', 'c', 'e']
print decisionTree.majorityCnt(classList)
# print '... 测试绘制树节点'
# plotDecisionTree.createPlot()
print '... 测试绘制决策树'
myTree = plotDecisionTree.retrieveTree(0)
leafNums = decisionTree.getNumLeafs(myTree)
treeDepth = decisionTree.getTreeDepth(myTree)
print myTree
print '叶子数量:%d, 树高度:%d' % (leafNums, treeDepth)
# plotDecisionTree.createPlot(myTree)
print '... 预测'
featNames = ['no surfacing', 'flippers', 'fish']
print decisionTree.classify(myTree, featNames, [1, 1])
print '... 测试和读取决策树存储'
decisionTree.storeTree(myTree, 'classfierStorage.txt')
print decisionTree.grabTree('classfierStorage.txt')
print '... 测试中文情况的决策树读取和存储'
cnTree = plotDecisionTree.retrieveTree(2)
print '存储前的决策树: \n', cnTree
# 对于中文来说,正常的print只能打印出utf-8编码格式
# 但是可以递归的打印字典 就可以输出中文
decisionTree.storeTree(cnTree, 'cnTree.txt')
print decisionTree.grabTree('cnTree.txt')
def demo():
print '... demo'
myDat, featNames = createDataSet()
print myDat
shannonEnt = decisionTree.calcShannonEnt(myDat)
print '当前数据集的熵是: ', shannonEnt
print '... 测试拆分数据集'
print decisionTree.splitDataSet(myDat, 0, 1)
print '... 测试最佳特征选择'
bestFeature = decisionTree.chooseBestFeatureToSplit(myDat)
print '最好的分类特征是 %s' % bestFeature
print '... 测试决策树的生成'
myTree = decisionTree.createTree(myDat, featNames)
print '生成的决策树是: \n', myTree
print '... 测试SortedCount'
classList = ['a', 'b', 'b', 'c', 'e']
print decisionTree.majorityCnt(classList)
# print '... 测试绘制树节点'
# plotDecisionTree.createPlot()
print '... 测试绘制决策树'
myTree = plotDecisionTree.retrieveTree(0)
leafNums = decisionTree.getNumLeafs(myTree)
treeDepth = decisionTree.getTreeDepth(myTree)
print myTree
print '叶子数量:%d, 树高度:%d' % (leafNums, treeDepth)
# plotDecisionTree.createPlot(myTree)
print '... 预测'
featNames = ['no surfacing', 'flippers', 'fish']
print decisionTree.classify(myTree, featNames, [1, 1])
print '... 测试和读取决策树存储'
decisionTree.storeTree(myTree, 'classfierStorage.txt')
print decisionTree.grabTree('classfierStorage.txt')
print '... 测试中文情况的决策树读取和存储'
cnTree = plotDecisionTree.retrieveTree(2)
print '存储前的决策树: \n', cnTree
# 对于中文来说,正常的print只能打印出utf-8编码格式
# 但是可以递归的打印字典 就可以输出中文
decisionTree.storeTree(cnTree, 'cnTree.txt')
print decisionTree.grabTree('cnTree.txt')
def test(inputTree, labels, data, prt=True):
'''
测试分类器效果
:param inputTree: 分类器
:param labels: 特征列
:param data: 二维list
:return: 准确率, 混淆矩阵
'''
count = 0
escape = 0
ac = 0
acpc = 0
au = 0
aupu = 0
for i in range(len(data)):
dataVec = list(data[i])
predict = classify(inputTree, labels, dataVec)
if predict == None:
escape += 1
continue
if dataVec[-1] == ' >50K':
au += 1
if predict == ' >50K':
aupu += 1
else:
ac += 1
if predict == ' <=50K':
acpc += 1
if predict == dataVec[-1]:
count += 1
accuracy = (count + escape) / float(len(data))
print('escape: {}'.format(escape))
confuse_matrix = np.array([[acpc, ac - acpc], [au - aupu, aupu]])
if prt:
print('accuracy: {}'.format(accuracy))
print(confuse_matrix)
return accuracy, confuse_matrix
def main():
dataSetX, dataSetY = readLensesData()
featureList = ['age', 'prescript', 'astigmatic', 'tearRate']
print(featureList)
threshold = 0
# create decisionTree
tree = decisionTree.createTree(dataSetX, dataSetY, featureList, threshold)
print(tree)
# save
import pickle
fo = open('tree.file', 'w+')
pickle.dump(tree, fo)
fo.close()
# load and use
inX = ['young', 'myope', 'yes', 'normal']
fi = open('tree.file')
tree = pickle.load(fi)
fi.close()
inY = decisionTree.classify(tree, featureList, inX)
print(inX)
print(inY)
def randomForestClassify(myForest, featLables, testVec):
classCount = {}
for tree in myForest:
firstStr = tree.keys()[0]
secondDict = tree[firstStr]
# 找到特征列在原始数据集的位置
featIndex = featLables.index(firstStr)
classLabel = None
for key in secondDict.keys():
if testVec[featIndex] == key:
if type(secondDict[key]).__name__ == 'dict':
classLabel = classify(secondDict[key], featLables, testVec)
else:
classLabel = secondDict[key]
if not classLabel: # 没有目标分支, 提前终止
classList = {}
searchTree(tree, classList)
classLabel = sorted(classList.items(),
key=lambda item: item[1],
reverse=True)[0][0]
classCount.setdefault(classLabel, 0)
classCount[classLabel] += 1
return sorted(classCount.items(), key=lambda item: item[1],
reverse=True)[0][0]
#!/usr/bin/env python
import decisionTree as dt
def createDataset():
dataset = [[1, 1], [1, 1], [1, 0], [0, 1], [0, 1]]
labels = ['yes', 'yes', 'no', 'no', 'no']
return (dataset, labels)
if __name__ == '__main__':
(dataset, labels) = createDataset()
labelName = ["no surface", "flipper"]
tree = dt.createTree(dataset, labels, labelName)
print(tree)
labelName = ["no surface", "flipper"]
label = dt.classify(tree, labelName, [0, 0])
print(label)
#!/usr/bin/env python import decisionTree as dt def createDataset() : dataset = [[1, 1], [1, 1], [1, 0], [0, 1], [0, 1]] labels = ['yes', 'yes', 'no', 'no', 'no'] return (dataset, labels) if __name__ == '__main__' : (dataset, labels) = createDataset() labelName = ["no surface", "flipper"] tree = dt.createTree(dataset, labels, labelName) print(tree) labelName = ["no surface", "flipper"] label = dt.classify(tree, labelName, [0,0]) print(label)