#!/usr/bin/python
# -*- coding:utf-8 -*-
import numpy as np
from numpy import *
import matplotlib.pyplot as plt
#处理数据函数
def loadDataSet():
dataMat=[]
labelMat=[]
fr=open('C:\\Users\\root\\Desktop\\2017machinelearning\\machinelearninginaction-master\\machinelearninginaction-master\\Ch05\\testSet.txt')
for line in fr.readlines():
lineArr=line.strip().split()
dataMat.append([1.0,float(lineArr[0]),float(lineArr[1])])
labelMat.append(int(lineArr[2]))
return dataMat,labelMat
def Sigmoid(x):
f=1.0/(1+exp(-x))
return f
#梯度函数
def gradAscent(dataMat,labelMat):
datamatri=mat(dataMat)
labelMatri=mat(labelMat).transpose() #转化为列向量
m,n=shape(datamatri)
alpha=0.001
maxCycles=500
weights=ones((n,1))
#循环迭代次数maxCycles
for i in range(maxCycles):
# print i
h=Sigmoid(datamatri*weights) #矩阵相乘 初始值
error=(labelMatri-h) #错误数\
weights=weights+alpha*datamatri.transpose()*error
return weights
#随机梯度函数
def stocgradAscent0(dataMat,labelMat):
# datamatri=mat(dataMat)
# labelMatri=mat(labelMat).transpose() #转化为列向量
m,n=shape(dataMat)
alpha=0.01
# maxCycles=500
weights=ones(n)
print n
# print weights
#循环迭代次数maxCycles
for i in range(m):
h=Sigmoid(sum(dataMat[i]*weights)) #矩阵相乘 初始值
error=labelMat[i]-h #错误数\
weights=weights+alpha*error*dataMat[i]
return weights
#改进的随机梯度函数
def stocGradAscent1(dataMatri,classLabels,numIter=150):
m,n=shape(dataMatri)
weights=ones(n)
for j in range(numIter):
dataIndex=range(m)
for i in range(m):
alpha=4/(1.0+j+i)+0.0001
randIndex=int(random.uniform(0,len(dataIndex)))
h=Sigmoid(sum(dataMatri[randIndex]*weights))
error=classLabels[randIndex]-h
weights=weights+alpha*error*dataMatri[randIndex]
del(dataIndex[randIndex])
return weights
#画图测试
def plotBestFit(weights):
# weights=wei.getA()
dataMat,labelMat=loadDataSet()
dataArr=array(dataMat)
# labelMat=array(labelMat)
# print type(labelMat[0][0])
n=shape(dataArr)[0] #取行数
xcord1=[]
xcord2 = []
ycord1=[]
ycord2 = []
for i in range(n):
if int(labelMat[i])==1:
xcord1.append(dataArr[i,1])
ycord1.append(dataArr[i,2])
# print int(labelMat[i][0]) == 1
else:
# print int ( labelMat[i][0] ) == 1
xcord2.append(dataArr[i,1])
ycord2.append(dataArr[i,2])
# weights = gradAscent(dataMat,labelMat)
fig=plt.figure()
ax=fig.add_subplot(111)
ax.scatter(xcord1,ycord1,c='red')
ax.scatter(xcord2,ycord2,c='g')
x=arange(-3.0,3.0,0.1)
y=(-weights[0]-weights[1]*x)/weights[2]
ax.plot(x,y.reshape(-1,1))
plt.show()
#梯度函数
a,b=loadDataSet()
weights=gradAscent(a,b)
plotBestFit(weights)
#随机梯度函数
a,b=loadDataSet()
weights=stocgradAscent0(array(a),b)
plotBestFit(weights)
##改进的随机梯度函数
a,b=loadDataSet()
weights=stocgradAscent0(array(a),b)
plotBestFit(weights)
#实例分析从疝气病病症预测病骂的死亡率
def classficatinon(Inter,weights):
a=sum(Inter*weights)
if Sigmoid(a)>0.5:
b=1
else:
b=0
return b
def colicTest():
frTrain=open('C:\\Users\\root\\Desktop\\2017machinelearning\\machinelearninginaction-master\\machinelearninginaction-master\\Ch05\\horseColicTraining.txt')
frTest=open('C:\\Users\\root\\Desktop\\2017machinelearning\\machinelearninginaction-master\\machinelearninginaction-master\\Ch05\\horseColicTest.txt')
trainingSet=[]
trainingLabels=[]
for line in frTrain.readlines():
currLine=line.strip().split('\t')
lineArr=[]#特征向量
for i in range(21):
lineArr.append(float(currLine[i]))
trainingSet.append(lineArr)
trainingLabels.append(float(currLine[21]))#训练集标签
trainWeiht=stocGradAscent1(array(trainingSet),trainingLabels,1)#训练出回归系数
errorCount=0
numTestVec=0.0
for line in frTest.readlines():
numTestVec+=1.0
currLine = line.strip ().split ( '\t' )
lineArr = [] # 特征向量(
for i in range ( 21 ):
lineArr.append ( float ( currLine[i] ) )
if int(classficatinon(array(lineArr),trainWeiht))!=int(currLine[21]):
errorCount+=1
errorRate=(float(errorCount)/numTestVec)
print "错误率为 %f" % errorRate
return errorRate
def multiTest():
numTests=1
errorSum=0.0
for k in range(numTests):
errorSum+=colicTest()
print "after %d 迭代后平均错误率为:%f" %(numTests,errorSum/float(numTests))
# colicTest()
multiTest()
#梯度函数
#随机梯度函数
##改进的随机梯度函数
#实例分析从疝气病病症预测病骂的死亡率
错误率为 0.288136
错误率为 0.300000
错误率为 0.295082
错误率为 0.290323
错误率为 0.285714
错误率为 0.281250
错误率为 0.276923
错误率为 0.287879
错误率为 0.298507
after 10 迭代后平均错误率为:0.338806