python实现决策树分类（2）
在上一篇文章中，我们已经构建了决策树，接下来可以使用它用于实际的数据分类。在执行数据分类时，需要决策时以及标签向量。程序比较测试数据和决策树上的数值，递归执行直到进入叶子节点。
这篇文章主要使用决策树分类器就行分类，数据集采用UCI数据库中的红酒，白酒数据，主要特征包括12个，主要有非挥发性酸,挥发性酸度,柠檬酸,残糖含量,氯化物,游离二氧化硫,总二氧化硫,密度,pH,硫酸盐,酒精,质量等特征。
下面是具体代码的实现：
#coding:utf-8
'''
2017.6.26author:Erin
function:"decesiontree"ID3
'''
importnumpyasnp
importpandasaspd
frommathimportlog
importoperator
importrandom
defload_data():
red=[line.strip().split(';')forlineinopen('e:/a/winequality-red.csv')]
white=[line.strip().split(';')forlineinopen('e:/a/winequality-white.csv')]
data=red+white
random.shuffle(data)#打乱data
x_train=data[:800]
x_test=data[800:]
features=['fixed','volatile','citric','residual','chlorides','free','total','density','pH','sulphates','alcohol','quality']
returnx_train,x_test,features
defcal_entropy(dataSet):
numEntries=len(dataSet)
labelCounts={}
forfeatVecindataSet:
label=featVec[-1]
iflabelnotinlabelCounts.keys():
labelCounts[label]=0
labelCounts[label]+=1
entropy=0.0
forkeyinlabelCounts.keys():
p_i=float(labelCounts[key]/numEntries)
entropy-=p_i*log(p_i,2)#log(x,10)表示以10为底的对数
returnentropy
defsplit_data(data,feature_index,value):
'''
划分数据集
feature_index：用于划分特征的列数，例如“年龄”
value:划分后的属性值：例如“青少年”
'''
data_split=[]#划分后的数据集
forfeatureindata:
iffeature[feature_index]==value:
reFeature=feature[:feature_index]
reFeature.extend(feature[feature_index+1:])
data_split.append(reFeature)
returndata_split
defchoose_best_to_split(data):
'''
根据每个特征的信息增益，选择最大的划分数据集的索引特征
'''
count_feature=len(data[0])-1#特征个数4
#print(count_feature)#4
entropy=cal_entropy(data)#原数据总的信息熵
#print(entropy)#0.9402859586706309
max_info_gain=0.0#信息增益最大
split_fea_index=-1#信息增益最大，对应的索引号
foriinrange(count_feature):
feature_list=[fe_index[i]forfe_indexindata]#获取该列所有特征值
#######################################
#print(feature_list)
unqval=set(feature_list)#去除重复
Pro_entropy=0.0#特征的熵
forvalueinunqval:#遍历改特征下的所有属性
sub_data=split_data(data,i,value)
pro=len(sub_data)/float(len(data))
Pro_entropy+=pro*cal_entropy(sub_data)
#print(Pro_entropy)
info_gain=entropy-Pro_entropy
if(info_gain>max_info_gain):
max_info_gain=info_gain
split_fea_index=i
returnsplit_fea_index
##################################################
defmost_occur_label(labels):
#sorted_label_count[0][0]次数最多的类标签
label_count={}
forlabelinlabels:
iflabelnotinlabel_count.keys():
label_count[label]=0
else:
label_count[label]+=1
sorted_label_count=sorted(label_count.items(),key=operator.itemgetter(1),reverse=True)
returnsorted_label_count[0][0]
defbuild_decesion_tree(dataSet,featnames):
'''
字典的键存放节点信息，分支及叶子节点存放值
'''
featname=featnames[:]################
classlist=[featvec[-1]forfeatvecindataSet]#此节点的分类情况
ifclasslist.count(classlist[0])==len(classlist):#全部属于一类
returnclasslist[0]
iflen(dataSet[0])==1:#分完了,没有属性了
returnVote(classlist)#少数服从多数
#选择一个最优特征进行划分
bestFeat=choose_best_to_split(dataSet)
bestFeatname=featname[bestFeat]
del(featname[bestFeat])#防止下标不准
DecisionTree={bestFeatname:{}}
#创建分支,先找出所有属性值,即分支数
allvalue=[vec[bestFeat]forvecindataSet]
specvalue=sorted(list(set(allvalue)))#使有一定顺序
forvinspecvalue:
copyfeatname=featname[:]
DecisionTree[bestFeatname][v]=build_decesion_tree(split_data(dataSet,bestFeat,v),copyfeatname)
returnDecisionTree
defclassify(Tree,featnames,X):
classLabel=''
root=list(Tree.keys())[0]
firstDict=Tree[root]
featindex=featnames.index(root)#根节点的属性下标
#classLabel='0'
forkeyinfirstDict.keys():#根属性的取值,取哪个就走往哪颗子树
ifX[featindex]==key:
iftype(firstDict[key])==type({}):
classLabel=classify(firstDict[key],featnames,X)
else:
classLabel=firstDict[key]
returnclassLabel
if__name__=='__main__':
x_train,x_test,features=load_data()
split_fea_index=choose_best_to_split(x_train)
newtree=build_decesion_tree(x_train,features)
#print(newtree)
#classLabel=classify(newtree,features,['7.4','0.66','0','1.8','0.075','13','40','0.9978','3.51','0.56','9.4','5'])
#print(classLabel)
count=0
fortestinx_test:
label=classify(newtree,features,test)
if(label==test[-1]):
count=count+1
acucy=float(count/len(x_test))
print(acucy)
测试的准确率大概在0.7左右。至此决策树分类算法结束。本文代码地址
以上就是本文的全部内容，希望对大家的学习有所帮助，也希望大家多多支持毛票票。