K-近邻算法
K-近邻算法的一般流程
from numpy import *
import operator
def createDataSet():
group=array([[1.0,1.1],[1.0,1.0],[0,0],[0,0.1]])
labels=['A','B','C','D']
return group,labels
def classify0(inX,dataSet,labels,k):
dataSetSize=dataSet.shape[0]
#距离计算
diffMat=tile(inX,(dataSetSize,1))-dataSet
sqDoffMat=diffMat**2
sqDistances=sqDoffMat.sum(axis=1)
distances=sqDistances**0.5
sortedDistIndices=distances.argsort()
classCount={}
#选择激励最小的K个点
for i in range(k):
voteIlabel=labels[sortedDistIndices[i]]
classCount[voteIlabel]=classCount.get(voteIlabel,0)+1
sortedClassCount=sorted(classCount.items(),
key=operator.itemgetter(1),reverse=True)
return sortedClassCount[0][0]
group,labels=createDataSet()
classify0([0,0],group,labels,3)
'C'
函数详解
tile([1,2],2)
array([1, 2, 1, 2])
tile([1,2],(2,2))
array([[1, 2, 1, 2],
[1, 2, 1, 2]])
x=array([[1,2,3],[2,3,4]])
print(x.shape)
print(x.shape[0])
(2, 3)
2
x=array([[1,2,3],[2,3,4]])
print(x**2)
[[ 1 4 9]
[ 4 9 16]]
x = np.array([[0, 3], [2, 2]])
np.argsort(x, axis=0)
np.argsort(x, axis=1)
array([[0, 1],
[1, 0]])
array([[0, 1],
[0, 1]])
dict = {'Name': 'Zara', 'Age': 27}
print "Value : %s" % dict.get('Age')
print "Value : %s" % dict.get('Sex', "Never")
Value : 27
Value : Never
""
Python 字典(Dictionary) items() 函数以列表返回可遍历的(键, 值) 元组数组
""
dict = {'Google': 'www.google.com', 'Runoob': 'www.runoob.com', 'taobao': 'www.taobao.com'}
print "字典值 : %s" % dict.items()
# 遍历字典列表
for key,values in dict.items():
print key,values
字典值 : [('Google', 'www.google.com'), ('taobao', 'www.taobao.com'), ('Runoob', 'www.runoob.com')]
Google www.google.com
taobao www.taobao.com
Runoob www.runoob.com
#operator库块提供了一系列的函数操作。比如,operator.add(x, y)等于x+y
abs(...)
abs(a) -- Same as abs(a).
and_(...)
and_(a, b) -- Same as a & b.
contains(...)
contains(a, b) -- Same as b in a (note reversed operands).
eq(...)
eq(a, b) -- Same as a==b.
operator模块提供的itemgetter函数用于获取对象的哪些维的数据,参数为一些序号。operator.itemgetter函数获取的不是值,而是定义了一个函数,通过该函数作用到对象上才能获取值。
a = [1,2,3]
>>> b=operator.itemgetter(1) //定义函数b,获取对象的第1个域的值
>>> b(a)
2
>>> b=operator.itemgetter(1,0) //定义函数b,获取对象的第1个域和第0个的值
>>> b(a)
(2, 1)
sorted函数用来排序,sorted(iterable[, cmp[, key[, reverse]]])
其中key的参数为一个函数或者lambda函数。所以itemgetter可以用来当key的参数
a = [(‘john’, ‘A’, 15), (‘jane’, ‘B’, 12), (‘dave’, ‘B’, 10)]
根据第二个域和第三个域进行排序
sorted(students, key=operator.itemgetter(1,2))
只要是可迭代对象都可以用sorted 。
sorted(itrearble, cmp=None, key=None, reverse=False)
=号后面是默认值 默认是升序排序的, 如果想让结果降序排列,用reverse=True
最后会将排序的结果放到一个新的列表中, 而不是对iterable本身进行修改。
1, 简单排序
sorted('123456') 字符串
['1', '2', '3', '4', '5', '6']
sorted([1,4,5,2,3,6]) 列表
[1, 2, 3, 4, 5, 6]
sorted({1:'q',3:'c',2:'g'}) 字典, 默认对字典的键进行排序
[1, 2, 3]
sorted({1:'q',3:'c',2:'g'}.keys()) 对字典的键
[1, 2, 3]
sorted({1:'q',3:'c',2:'g'}.values()) 对字典的值
['c', 'g', 'q']
sorted({1:'q',3:'c',2:'g'}.items()) 对键值对组成的元组的列表
[(1, 'q'), (2, 'g'), (3, 'c')]
自定义比较函数
def comp(x, y):
if x < y:
return 1
elif x > y:
return -1
else:
return 0
nums = [3, 2, 8 ,0 , 1]
nums.sort(comp)
print nums # 降序排序[8, 3, 2, 1, 0]
nums.sort(cmp) # 调用内建函数cmp ,升序排序
print nums # 降序排序[0, 1, 2, 3, 8]
key在使用时必须提供一个排序过程总调用的函数
x = ['mmm', 'mm', 'mm', 'm' ]
x.sort(key = len)
print x # ['m', 'mm', 'mm', 'mmm']
在约会网站上使用K近邻算法
使用算法:产生简单的命令行程序,然后海伦可以输入一些特征数据以判断对方是否为自己喜欢的类型
完整代码:
from numpy import *
import operator
def classify0(inX,dataSet,labels,k):
dataSetSize=dataSet.shape[0]
#距离计算
diffMat=tile(inX,(dataSetSize,1))-dataSet
sqDoffMat=diffMat**2
sqDistances=sqDoffMat.sum(axis=1)
distances=sqDistances**0.5
sortedDistIndices=distances.argsort()
classCount={}
#选择激励最小的K个点
for i in range(k):
voteIlabel=labels[sortedDistIndices[i]]
classCount[voteIlabel]=classCount.get(voteIlabel,0)+1
sortedClassCount=sorted(classCount.items(),
key=operator.itemgetter(1),reverse=True)
return sortedClassCount[0][0]
def file2matrix(filename):
fr=open(filename)
arrayOLines=fr.readlines()
numberOfLines=len(arrayOLines)
returnMat=zeros((numberOfLines,3))
classLabelVector=[]
index=0
for line in arrayOLines:
#跳过所有的空格字符,使用tab‘\t’分割数据
line=line.strip()
listFromLine=line.split('\t')
returnMat[index,:]=listFromLine[0:3]
classLabelVector.append(int(listFromLine[-1]))
index+=1
return returnMat,classLabelVector
data_path='E:/dataset/machinelearninginaction/Ch02/'
datMat,datLabel=file2matrix(data_path+'datingTestSet2.txt')
print(datMat)
print(datLabel[0:20])
#分析数据:使用matplotlib创建散点图
import matplotlib
import matplotlib.pyplot as plt
fig=plt.figure()
ax=fig.add_subplot(111)
ax.scatter(datMat[:,0],datMat[:,1],
15.0*array(datLabel),15.0*array(datLabel))
plt.show()
#准备数据:归一化数据
def autoNorm(dataSet):
minVals=dataSet.min(0)
maxVals=dataSet.max(0)
ranges=maxVals-minVals
m=dataSet.shape[0]
normData=dataSet-tile(minVals,(m,1))
normData=normData/tile(ranges,(m,1))
return normData,ranges,minVals
normData,ranges,minVals=autoNorm(datMat)
print(normData)
print(ranges)
#测试算法:作为完整程序验证分类器
def datingClassTest():
hoRatio=0.10
datingDataMat,datingDatalabel=file2matrix(data_path+\
'datingTestSet2.txt')
norm,range1,minVals=autoNorm(datingDataMat)
m=norm.shape[0]
numTest=int(m*hoRatio)
errorCount=0
for i in range(numTest):
classResult=classify0(norm[i,:],norm[numTest:m,:],\
datLabel[numTest:m],3)
print('分类器学习的结果,%d,真实值是%d'%(classResult,datingDatalabel[i]))
if(classResult!=datingDatalabel[i]):errorCount+=1
print("错误率是:%f"%(errorCount/numTest))
if __name__ == '__main__':
datingClassTest()
实例:手写识别系统
准备数据:将图像转换为测试向量
from numpy import *
import operator
def img2vector(filename):
file_path = "E:/dataset/machinelearninginaction/Ch02/digits/trainingDigits/"
returnVec=zeros((1,1024))
fr=open(file_path+filename)
for i2 in range(32):
lineStr=fr.readline()
for j2 in range(32):
returnVec[0,32*i2+j2]=int(lineStr[j2])
return returnVec
def classify0(inX,dataSet,labels,k):
dataSetSize=dataSet.shape[0]
#距离计算
diffMat=tile(inX,(dataSetSize,1))-dataSet
sqDoffMat=diffMat**2
sqDistances=sqDoffMat.sum(axis=1)
distances=sqDistances**0.5
sortedDistIndices=distances.argsort()
classCount={}
#选择激励最小的K个点
for i in range(k):
voteIlabel=labels[sortedDistIndices[i]]
classCount[voteIlabel]=classCount.get(voteIlabel,0)+1
sortedClassCount=sorted(classCount.items(),
key=operator.itemgetter(1),reverse=True)
return sortedClassCount[0][0]
import os
def handWritingClassTest():
hwLabels=[]
trainingFileList=os.listdir('E:/dataset/machinelearninginaction/Ch02/digits/trainingDigits')
m=len(trainingFileList)
print(m)
trainVec=zeros((m,1024))
#traingLabel=zeros((m,1))
traingLabel=[]
i=0
for filename in trainingFileList:
img=img2vector(filename)
trainVec[i,:]=img
label=filename.split('_')[0]
traingLabel.append(int(label))
i+=1
testFileList=os.listdir("E:/dataset/machinelearninginaction/Ch02/digits/testDigits")
#n=len(testFileList)
i=0
for filename in testFileList:
img=img2vector(filename)
#testVect[i]=img
resu=classify0(img,trainVec,traingLabel,3)
label = filename.split('_')[0]
print("predict:%d the true value:%d"%(resu,int(label)))
if(resu!=label):
i+=1
print("the precision is %f"%(i/len(testFileList)))
file_path="E:/dataset/machinelearninginaction/Ch02/digits/trainingDigits/"
#testVec=img2vector(file_path+"0_0.txt")
#print(testVec[0:32])
handWritingClassTest()
the precision is 1.000000