决策树

一、主要步骤

特征选择

决策树的生成

决策树的修剪

二、算法实现

'''
创建分支的伪代码函数creatBranch():
    
检测数据集中的每个子项是否属于同一分类：
    if so return 类标签；
    Else 
        寻找划分数据集的最好特征
        划分数据集
        创建分支节点
            for 每个划分的子集
                调用函数createBranch并增加返回结果到分支节点中
        return 分支节点
'''

def createDataSet():
    dataSet = [[1, 1, 'yes'],
               [1, 1, 'yes'],
               [1, 0, 'no'],
               [0, 1, 'no'],
               [0, 1, 'no']]
    labels = ['no surfacing','flippers']
    return dataSet, labels
dataSet, labels = createDataSet()

from math import log
def calcShannonEnt(dataSet): 
    '''
    功能：
    计算给定数集的香农熵
    '''
    numEntries = len(dataSet)
    labelCounts = {}
    for featVec in dataSet: 
        currentLabel = featVec[-1]
        if currentLabel not in labelCounts.keys(): labelCounts[currentLabel] = 0
        labelCounts[currentLabel] += 1
    shannonEnt = 0.0
    for key in labelCounts:
        prob = float(labelCounts[key])/numEntries
        shannonEnt -= prob * log(prob,2) 
    return shannonEnt

shannonEnt = calcShannonEnt(dataSet)

'''
3.2 划分数据集
'''
def splitDataSet(dataSet, axis, value):
    '''
    功能：按照给定特征划分数据
    参数：
        dataSet：待划分的数据集
        axis：特征值所在特征集的列
        value：特征值     
    '''
    retDataSet = []
    for featVec in dataSet:
        if featVec[axis] == value:
            reducedFeatVec = featVec[:axis]     
            reducedFeatVec.extend(featVec[axis+1:])
            retDataSet.append(reducedFeatVec)
    return retDataSet

retDataSet = splitDataSet(dataSet, 0, 1)

def chooseBestFeatureToSplit(dataSet):
    '''
    功能：选取特征，划分数据集，计算得出最好的划分数据集的特征
    
    '''
    numFeatures = len(dataSet[0]) - 1        #特征集长度 
    baseEntropy = calcShannonEnt(dataSet)    #整个数据集的原始香浓熵
    bestInfoGain = 0.0; bestFeature = -1     
    for i in range(numFeatures):        
        featList = [example[i] for example in dataSet]
        uniqueVals = set(featList)                  #得到某个特征的所有取值
        newEntropy = 0.0
        for value in uniqueVals:                    #计算条件熵
            subDataSet = splitDataSet(dataSet, i, value)
            prob = len(subDataSet)/float(len(dataSet))
            newEntropy += prob * calcShannonEnt(subDataSet)     
        infoGain = baseEntropy - newEntropy         #得到信息增益
        if (infoGain > bestInfoGain):       
            bestInfoGain = infoGain         
            bestFeature = i
    return bestFeature                      

bestFeature = chooseBestFeatureToSplit(dataSet)

'''
3.3 递归构建决策树
'''
import operator

def createTree(dataSet,labels):
    '''
    功能：使用递归的方法构建二叉树
    参数：
        dataSet：数据集
        labels：标签列表
    
    '''
    classList = [example[-1] for example in dataSet]
    if classList.count(classList[0]) == len(classList): #判断结果是否全部相同
        return classList[0]
    if len(dataSet[0]) == 1:                            
        return majorityCnt(classList)
    bestFeat = chooseBestFeatureToSplit(dataSet)        #得到最好的划分数据集的特征
    bestFeatLabel = labels[bestFeat]
    myTree = {bestFeatLabel:{}}
    del(labels[bestFeat])                               
    featValues = [example[bestFeat] for example in dataSet]
    uniqueVals = set(featValues)
    for value in uniqueVals:
        subLabels = labels[:]       
        myTree[bestFeatLabel][value] = createTree(splitDataSet(dataSet, bestFeat, value),subLabels)
    return myTree    

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]

myTree = createTree(dataSet,labels)