机器学习实战K-近邻算法
BonjourFFF 2017-10-07
今天开始学习机器学习，第一章是K-近邻算法，有不对的地方请指正
大概总结一下近邻算法写分类器步骤：
1. 计算测试数据与已知数据的特征值的距离，离得越近越相似
2. 取距离最近的K个已知数据的所属分类
3. 最后统计K个值的分类分别出现的概率，返回最多的一个属性，即为测试数据的所属分类
4. 至于怎么把文本转换成numpy的类型，需要学习numpy模块的相关知识，附上
numpy学习连接 http://old.sebug.net/paper/books/scipydoc/numpy_intro.html
#-*- coding:utf-8 *-*-
from numpy import *
import operator #计算模块
import matplotlib
import matplotlib.pyplot as plt
import time
import random
from mpl_toolkits.mplot3d import Axes3D
from os import listdir
import time
def createDataSet():
group = array([[1.0,1.1],[1.0,1.0],[0,0],[0,0.1]])
labels = ['A','A','B','B']
return group,labels
#A,B分类
def classify0(inX, dataSet, labels, k):
dataSetSize = dataSet.shape[0]
diffMat = tile(inX,(dataSetSize,1)) - dataSet #tile函数把inx复制datasetsize行1列
sqDiffMat = diffMat**2
#print "sqDiffMat : ",sqDiffMat
sqDistance = sqDiffMat.sum(axis = 1)
distance = sqDistance**0.5
#print "distance : ",distance
sortedDistIndicies = distance.argsort() #返回从小到大的元素的下标，比如[1 3 2 4].argsort()返回[0 2 1 3]
#print "****",sortedDistIndicies
classCount = {}
for i in range(k):
voteIlabel = labels[sortedDistIndicies[i]] #统计各个现有值所属的特征向量
#print sortedDistIndicies[i],voteIlabel
classCount[voteIlabel] = classCount.get(voteIlabel,0)+1 #统计各个特征向量出现的次数
sortedClassCount = sorted(classCount.iteritems(),key = operator.itemgetter(1),reverse = True)
#operator.itemgetter()从小到大排序
#print "sortedClassCount : ",sortedClassCount
return sortedClassCount[0][0]
group,labels = createDataSet()
#print classify0([0,0], group, labels, 3)
# # a = [('b',2),('a',1),('c',0)]
# a=[('b',2),('a',2),('a',1),('c',0)]
# b = sorted(a,key = operator.itemgetter(0)) #优先根据第一个元素排序
# print b
# b = sorted(a,key = operator.itemgetter(1)) #优先根据第二个元素排序
# print b
# b = sorted(a,key = operator.itemgetter(1,0)) #优先根据第二个元素排序，当第二个元素相等的情况下根据第一个元素排序
# print b
#解析数据
def file2matrix(filename):
with open(filename) as f:
lines = f.readlines()
matrixNumber = len(lines)
print 'the all lines is :',matrixNumber
#matrix = zeros((matrixNumber,3),dtype = 'int') #生成空的n行3列的矩阵
matrix = zeros((matrixNumber,2))
vector = []
index = 0 #矩阵索引
for line in lines:
line = line.strip()
data = line.split("\t")
matrix[index:] = data[0:2] #把提取出来的复制到矩阵里面
vector.append(int((data[-1]))) #最后一个特征值作为特征向量
index+=1
return matrix,vector
#生成文本数据
def createdata(filename):
with open(filename,'w') as f:
for i in range(1000):
r1 = int(random.random()*1000)
r2 = 0
if(0<=r1<=200):
r2 = 1
if(200<r1<=400):
r2 = 2
if(400<r1<=600):
r2 = 3
if(600<r1<=800):
r2 = 4
if(800<r1<=1000):
r2 = 5
r1 = str(r1)
r2 = str(r2)
#r2 = str(int(random.random()*10))
r3 = str(int(random.random()*10))
f.writelines(r3+'\t'+r1+'\t'+r2+'\n')
#createdata(r'D:\test_packages\knntest.txt')
'''
datat,labels = file2matrix(r'D:\test_packages\knntest.txt')
print datat
# print datat[:,1] #纵向的第二列
# print datat[:][1] #横向的第二列
print labels
fig = plt.figure() #生成容器
plt.title('favorite table data')
ax = fig.add_subplot(1,1,1,projection='3d') #3D模型
ax.scatter(datat[:,0],datat[:,1],datat[:,2],array(labels),array(labels),array(labels)) #使用datat的第二列和第三列作为X轴和Y轴的值
ax.legend()
plt.show()
fig = plt.figure()
ax = fig.add_subplot(1,1,1) #把容器划分为1行1列，图像画在第一格,背景颜色为axisbg = ‘’
ax.scatter(datat[:,1],datat[:,2],array(labels),array(labels)) #使用datat的第二列和第三列作为X轴和Y轴的值
#ax.grid(True) #是否显示网格
# plt.show()
plt.show()
'''
#归一化，(old-min)/(max-min)
def autoNormal(dataSet):
maxVals = dataSet.max(0) #纵向找到每一个样本的最大特征值
minVals = dataSet.min(0)
ranges = maxVals - minVals #计算差值
normalValue = zeros(shape(dataSet))
m = dataSet.shape[0]
normalValue = dataSet - tile(minVals,(m,1)) #计算(old-min)
normalValue = normalValue/tile(ranges,(m,1))
return normalValue,ranges,minVals
#归一化特征值之后
datat,labels = file2matrix(r'D:\test_packages\knntest.txt')
normalValue,ranges,minVals = autoNormal(datat)
print normalValue
fig = plt.figure()
ax = fig.add_subplot(1,1,1) #把容器划分为1行1列，图像画在第一格,背景颜色为axisbg = ‘’
ax.scatter(normalValue[:,0],normalValue[:,1],array(labels),array(labels)) #使用datat的第二列和第三列作为X轴和Y轴的值
#ax.grid(True) #是否显示网格
# plt.show()
plt.show()
#约会网站测试函数
def datinggTest():
datat,labels = file2matrix(r'D:\test_packages\knntest.txt')
normal,ranges,minvals = autoNormal(datat)
testData = 0.5 #10%用来测试，90%用来训练
testNumber = normal.shape[0] #总行数
numberTestValues = int(testNumber*testData) #测试行数
error = 0.0
for i in range(numberTestValues):
labelValue = classify0(normal[i,:], normal[numberTestValues:testNumber,:], labels[numberTestValues:testNumber], 3)
if (labelValue != labels[i]):
error+=1.0
print "this time is error the error is %s, the right is %s"%(labelValue,labels[i])
else:
print "all right ,the number is %s, the right is %s"%(labelValue,labels[i])
error_result = ((error/float(numberTestValues)))
print "your error_result is %s"%(error_result)
print 'error is :',error
datinggTest()
#把二进制文件转化为np.array
def img2Vector(filename):
with open(filename) as f:
vector = zeros((1,1024))
for i in range(32):
line = f.readline()
for j in range(32):
vector[0,32*i+j] = line[j]
return vector
vector = img2Vector(r'D:\test_packages\trainingDigits\0_0.txt')
print vector[0,11:17]
#手写数字识别系统测试代码
def handwritingClassTest():
startTime = time.ctime()
handLabels = []
trainFile = listdir(r'D:\test_packages\trainingDigits')
m = len(trainFile)
trainMat = zeros((m,1024))
for i in range(m):
fileName = trainFile[i]
file = fileName.split('.')[0]
classNumber = file.split('_')[0]
handLabels.append(classNumber)
trainMat[i,:] = img2Vector(r'D:\test_packages\trainingDigits\%s'%fileName)
testFiles = listdir(r'D:\test_packages\testDigits')
nTest = len(testFiles)
error = 0.0
for i in range(nTest):
fileName = testFiles[i]
file = fileName.split('.')[0]
classNumber = file.split('_')[0]
testMat = img2Vector(r'D:\test_packages\testDigits\%s'%fileName)
testLabels = classify0(testMat, trainMat, handLabels, 3)
if (testLabels != classNumber):
error+=1.0
print 'error , error number is %s, the right number is %s'%(testLabels,classNumber)
else:
print 'right'
error = error/float(nTest)
stopTime = time.ctime()
print 'all right ,the error_result is %s'%(error)
print 'the process start at %s'%(startTime)
print 'the process stop at %s'%(stopTime)
handwritingClassTest()
