机器学习-决策树算法代码详解

from imp import reload
from math import log
import operator

def createDataSet():
    dataSet = [[1, 1, 'yes'],
               [1, 1, 'yes'],
               [1, 0, 'no'],
               [0, 1, 'no'],
               [0, 1, 'no']]
    labels = ['no surfacing','flippers']
    #change to discrete values
    return dataSet, labels

def calcShannonEnt(dataSet):#dateset is a n*m
    numEntries = len(dataSet)#the row of the dateset,n
    labelCounts = {}#this is a dict,used to count the lables
    for featVec in dataSet: #the the number of unique elements and their occurance
        currentLabel = featVec[-1]#featVec is a list ,it is the row of the dateset
        if currentLabel not in labelCounts.keys(): labelCounts[currentLabel] = 0
        labelCounts[currentLabel] += 1
    shannonEnt = 0.0
    for key in labelCounts:#key is the KEY,and lableCounts[key] is its value
        prob = float(labelCounts[key])/numEntries#统计所有的不同标签出现的频次，计算他出现的概率
        shannonEnt -= prob * log(prob,2) #log base 2，count shannnonEnt
    return shannonEnt
    
def splitDataSet(dataSet, axis, value):
    retDataSet = []#this is an empty list
    for featVec in dataSet:#featVec is the row of the dateset
        if featVec[axis] == value:#axis is its place where you want to split
            reducedFeatVec = featVec[:axis]     #chop out axis used for splitting
            reducedFeatVec.extend(featVec[axis+1:])#[1,2,3]extend[4,5]=[1,2,3,4,5]
            retDataSet.append(reducedFeatVec)#[1,2,3]append[4,5]=[1,2,3,[4,5]]
    return retDataSet
    
def chooseBestFeatureToSplit(dataSet):
    numFeatures = len(dataSet[0]) - 1      #the last column is used for the labels.the value of len(dataSet[0])is its number of collos,the value of len(dateset))is its number of rows
    baseEntropy = calcShannonEnt(dataSet)#count the original shannonEnt
    bestInfoGain = 0.0; bestFeature = -1
    for i in range(numFeatures):        #iterate over all the features，每行有多少列，即有多少个分类特征，遍历所有分类特征，选取划分后ShannonEnt最小的特征为划分结果
        featList = [example[i] for example in dataSet]#create a list of all the examples of this feature,example include many row-list of the dataSet,example[i] is the ith value of the dataSet
        uniqueVals = set(featList)       #get a set of unique values#$# alist=[example[i] for example in dataSet],you can get a total column of dataSet
        newEntropy = 0.0
        for value in uniqueVals:#计算当前列出现所有结果分割下的ShannonEnt
            subDataSet = splitDataSet(dataSet, i, value)#利用dataSet的每一列中出现的所有的uniqueVals分割dataset，计算概率和shannonEnt,并求和
            prob = len(subDataSet)/float(len(dataSet))
            newEntropy += prob * calcShannonEnt(subDataSet)     
        infoGain = baseEntropy - newEntropy     #calculate the info gain; ie reduction in entropy
        if (infoGain > bestInfoGain):       #compare this to the best gain so far
            bestInfoGain = infoGain         #if better than current best, set to best
            bestFeature = i                 #the ith feature is the better feature for now
    return bestFeature                      #returns an integer

def majorityCnt(classList):
    classCount={}
    for vote in classList:
        if vote not in classCount.keys(): classCount[vote] = 0
        classCount[vote] += 1
    sortedClassCount = sorted(classCount.iteritems(), key=operator.itemgetter(1), reverse=True)
    return sortedClassCount[0][0]

def createTree(dataSet,labels):
    classList = [example[-1] for example in dataSet]
    if classList.count(classList[0]) == len(classList): #第一个停止条件
        return classList[0]#stop splitting when all of the classes are equal
    if len(dataSet[0]) == 1: #stop splitting when there are no more features in dataSet
        return majorityCnt(classList)
    bestFeat = chooseBestFeatureToSplit(dataSet)#the return is an Int,which is a place of the row of the column`s feature，The best feature
    bestFeatLabel = labels[bestFeat]#特征中的一个
    myTree = {bestFeatLabel:{}}#创建一个树的字典
    del(labels[bestFeat])
    featValues = [example[bestFeat] for example in dataSet]#featValues是得到当前这一列特征的所有表达
    uniqueVals = set(featValues)
    for value in uniqueVals:
        subLabels = labels[:]       #copy all of labels, so trees don't mess up existing labels，在python中，参数是列表类型时，是按照引用方式传递，会改变他的值
        myTree[bestFeatLabel][value] = createTree(splitDataSet(dataSet, bestFeat, value),subLabels)
        #value的值是上字典key bestFeatLable对应的值，但同时value也是一个字典，value是该字典的key，等号右边的部分是key下的值，该值同时可能又是一个字典
    return myTree
'''
            bestFeatLabel='sunshine'
            myTree = {bestFeatLabel:{}}
            uniqueVals=['rain','mud','wind']
            for value in uniqueVals:
            myTree[bestFeatLabel][value]={"a dog"}
            print(myTree)
            >>{'sunshine': {'rain': {'a dog'}, 'mud': {'a dog'}, 'wind': {'a dog'}}}
                    
'''

    
def classify(inputTree,featLabels,testVec):#inputtree是已经生成了的树，testvec是测试向量，featlable是测试向量对应位置的标签
    firstStr = inputTree.keys()[0]
    secondDict = inputTree[firstStr]
    featIndex = featLabels.index(firstStr)
    key = testVec[featIndex]#判断测试向量在该分类下的值是否是决策树的叶节点，如果不是则继续递归判断，如果是，返回分类结果
    valueOfFeat = secondDict[key]
    if isinstance(valueOfFeat, dict): 
        classLabel = classify(valueOfFeat, featLabels, testVec)
    else: classLabel = valueOfFeat
    return classLabel