def testDigits(kTup=('rbf', 10)):
dataArr, labelArr = loadImages('trainingDigits')
b, alphas = svmMLiA.smoP(dataArr, labelArr, 200, 0.0001, 10000, kTup)
datMat = mat(dataArr)
labelMat = mat(labelArr).transpose()
svInd = nonzero(alphas.A > 0)[0]
sVs = datMat[svInd]
labelSV = labelMat[svInd]
print("there are %d Support Vectors" % shape(sVs)[0])
m, n = shape(datMat)
errorCount = 0
for i in range(m):
kernelEval = svmMLiA.kernelTrans(sVs, datMat[i, :], kTup)
predict = kernelEval.T * multiply(labelSV, alphas[svInd]) + b
if sign(predict) != sign(labelArr[i]): errorCount += 1
print("the training error rate is: %f" % (float(errorCount) / m))
dataArr, labelArr = loadImages('testDigits')
errorCount = 0
datMat = mat(dataArr)
labelMat = mat(labelArr).transpose()
m, n = shape(datMat)
for i in range(m):
kernelEval = svmMLiA.kernelTrans(sVs, datMat[i, :], kTup)
predict = kernelEval.T * multiply(labelSV, alphas[svInd]) + b
if sign(predict) != sign(labelArr[i]): errorCount += 1
print("the test error rate is: %f" % (float(errorCount) / m))
def testRbf(k1=1.3):
dataArr, labelArr = svmMLiA.loadDataSet('testSetRBF.txt')
b, alphas = svmMLiA.smoP(dataArr, labelArr, 200, 0.0001, 10000, ('rbf', k1))
dataMat = np.mat(dataArr)
labelMat = np.mat(labelArr).transpose()
svInd = np.nonzero(alphas.A > 0)[0]
sVs = dataMat[svInd]
labelSv = labelMat[svInd]
print('there are %d Support Vectors' % np.shape(sVs)[0])
m,n = np.shape(dataMat)
errorCount = 0
for i in range(m):
kernelEval = svmMLiA.kernelTrans(sVs, dataMat[i,:], ('rbf', k1))
predict = kernelEval.T * np.multiply(labelSv, alphas[svInd]) + b
if np.sign(predict) != np.sign(labelArr[i]):
errorCount += 1
print('the training error rate is: %f' % (float(errorCount) / m))
dataArr, labelArr = svmMLiA.loadDataSet('testSetRBF2.txt')
errorCount = 0
dataMat = np.mat(dataArr)
labelMat = np.mat(labelArr).transpose()
m,n = np.shape(dataMat)
for i in range(m):
kernelEval = svmMLiA.kernelTrans(sVs, dataMat[i,:], ('rbf', k1))
predict = kernelEval.T * np.multiply(labelSv, alphas[svInd]) + b
if np.sign(predict) != np.sign(labelArr[i]):
errorCount += 1
print('the test error rate is: %f' % (float(errorCount) / m))
def testRbf(k1=100):
dataArr,labelArr = svm.loadDataSet('../data/test1.txt')
dataMat = mat(dataArr)
#print dataMat
b,alphas = svm.smoP(dataArr, labelArr, 200, 0.01, 10000, ('rbf', k1)) #C=200 important
datMat=mat(dataArr); labelMat = mat(labelArr).transpose()
#print alphas.A>0
svInd=nonzero(alphas.A>0)[0]
sVs=datMat[svInd] #get matrix of only support vectors
#print dataMat
#ax.scatter(sVs[:,0], sVs[:,1], s=30, c='green', marker='s')
#ax.scatter(sVs[:,0], sVs[:,1])
labelSV = labelMat[svInd];
print "there are %d Support Vectors" % shape(sVs)[0]
m,n = shape(datMat)
errorCount = 0
for i in range(m):
kernelEval = svm.kernelTrans(sVs,datMat[i,:],('rbf', k1))
predict=kernelEval.T * multiply(labelSV,alphas[svInd]) + b
if sign(predict)!=sign(labelArr[i]): errorCount += 1
print "the training error rate is: %f" % (float(errorCount)/m)
dataArr,labelArr = svm.loadDataSet('../data/test2.txt')
errorCount = 0
datMat=mat(dataArr); labelMat = mat(labelArr).transpose()
m,n = shape(datMat)
fig = plt.figure()
ax = fig.add_subplot(111)
#print sVs
for i in range(m):
kernelEval = svm.kernelTrans(sVs,datMat[i,:],('rbf', k1))
predict=kernelEval.T * multiply(labelSV,alphas[svInd]) + b
if sign(predict)!=sign(labelArr[i]):
errorCount += 1
if sign(predict) > 0:
#c = [int((0.2/float(predict))) for cc in range(0, 60)]
#ax.scatter(range(0, 60), array(datMat[i,:]), array(c), array(c))
ax.scatter(range(0, 60), array(datMat[i,:]), s=20, c='green', marker='s')
#print datMat[i,:], sign(predict), sign(labelArr[i])
continue
else:
ax.scatter(range(0, 60), array(datMat[i,:]), s=20, c='red', marker='s')
#print datMat[i,:], sign(predict), sign(labelArr[i])
continue
for i in range(shape(sVs)[0]):
ax.scatter(range(0, 60), array(sVs[i]), s=10, c='yellow', marker='s')
continue
plt.show()
print "the test error rate is: %f" % (float(errorCount)/m)
def titanicBysvm(filename,k1=1.3):
trainingdataarr,traininglabelarr,testdataarr,testlaeblarr = loadDataSet(filename)
# C = 200 , toler = 0.0001 , maxiteration = 10000
# b,alphas = svm.smoSimple(trainingdataarr,traininglabelarr,100,0.001,10000)
b,alphas = svm.smoP(trainingdataarr,traininglabelarr, 200, 0.0001, 10000, ('rbf', k1))
# w f(x) = w.T*x + b
w = svm.calcWs(alphas,trainingdataarr,traininglabelarr)
dataMat = mat(trainingdataarr); labelmat = mat(traininglabelarr).transpose()
# svInd is the index of support vector,sVs is the support vectors,labelSV is the support vector labels
svInd = nonzero(alphas.A > 0)[0]
sVs = dataMat[svInd]
labelSV = labelmat[svInd]
print "there are %d Support Vectors" % shape(sVs)[0]
m,n = shape(trainingdataarr)
errorCount = 0
for i in range(m):
kernelEval = svm.kernelTrans(sVs,dataMat[i,:], ('rbf',k1))
predict = kernelEval.T * multiply(labelSV,alphas[svInd]) + b
# predict = trainingdataarr[i,:] * mat(w) + b
# print predict
# print trainingdataarr[i,:]
# print " "
# print predict
# print "\n"
if sign(predict) != sign(traininglabelarr[i]):
errorCount += 1
print "the number of errors is %d" % errorCount
print "the training accuracy rate is: %f" % (1 - (float)(errorCount) / m)
# test data running
testErrorCount = 0
dataMat = mat(testdataarr); testlabelMat = mat(testlaeblarr).transpose()
m,n = shape(testdataarr)
for i in range(m):
kernelEval = svm.kernelTrans(sVs,dataMat[i,:],('rbf',k1))
predict = kernelEval.T * multiply(labelSV,alphas[svInd]) + b
if sign(predict) != sign(testlaeblarr[i]):
testErrorCount += 1
print "the test accuracy rate is %.3f" % (1 - float(testErrorCount) / m)
import svmMLiA
from numpy import *
dataArr, labelArr = svmMLiA.loadDataSet('testSet.txt')
b, alphas = svmMLiA.smoP(dataArr, labelArr, 0.6, 0.001, 40)
ws = svmMLiA.calcWs(alphas, dataArr, labelArr)
print ws
datMat = mat(dataArr)
print datMat[0]*mat(ws)+b
print labelArr[0]
print datMat[1]*mat(ws)+b
print labelArr[1]
print datMat[2]*mat(ws)+b
print labelArr[2]
from numpy import *
import svmMLiA
dataArr, labelArr = svmMLiA.loadDataSet('testSet.txt')
b, alphas = svmMLiA.smoP(dataArr, labelArr, 0.6, 0.001, 40)
ws = svmMLiA.calcWs(alphas, dataArr, labelArr)
print(ws)
dataMat = mat(dataArr)
print(dataMat[0] * mat(ws) + b)
print(labelArr[0])
print(dataMat[2] * mat(ws) + b)
print(labelArr[2])
print(dataMat[1] * mat(ws) + b)
print(labelArr[1])
iter: 30 i:17, pairs changed 1
j not moving enough
iteration number: 0
j not moving enough
iteration number: 1
b: matrix([[-3.8486163]])
alphas[alphas>0] = matrix([[0.09313378, 0.27456007, 0.04445935, 0.3232345 ]])
"""
# To see which points of our dataset are support vectors
for i in range(100):
if alphas[i] > 0.0:
print(dataArr[i], labelArr[i])
"""
output:
[4.658191, 3.507396] -1.0
[3.457096, -0.082216] -1.0
[2.893743, -1.643468] -1.0
[6.080573, 0.418886] 1.0
"""
""" optimised SMOP """
b, alphas = SVM.smoP(dataArr, labelArr, 0.6, 0.001, 40)
""" get alpha for supporting vectors """
ws = SVM.calcWs(alphas, dataArr, labelArr)
""" test rbf kernel """
SVM.testRbf()
""" testdigits classification using rbf kernel """
SVM.testDigits(('rbf', 20))
import svmMLiA
dm, ls = svmMLiA.loadDataSet('testSet.txt')
b, alphas = svmMLiA.smoSimple(dm, ls, 0.6, 0.001, 40) # 简版smo算法
alphas[alphas > 0]
shape(alphas[alphas > 0])
svMat = []
for i in range(100):
if alphas[i] > 0.0:
svMat.append(dm[i])
print dm[i], ls[i] # 支持向量
svmMLiA.plot(dm, ls, svMat)
import svmMLiA
dm, ls = svmMLiA.loadDataSet('testSet.txt')
b, alphas = svmMLiA.smoP(dm, ls, 0.6, 0.001, 40)
ws = svmMLiA.calcWs(alphas, dm, ls)
mat(dm)[0] * mat(ws) + b # 对第0个数据分类
ls[0]
import svmMLiA
dm, ls = svmMLiA.loadDataSet('testSet.txt')
b, alphas = svmMLiA.smoP(dm, ls, 0.6, 0.001, 40)
svMat = []
for i in range(100):
if alphas[i] > 0.0:
svMat.append(dm[i])
ws = svmMLiA.calcWs(alphas, dm, ls)
svmMLiA.plot(dm, ls, svMat, ws, float(b))
#coding=utf-8
import svmMLiA as svm
import functionsCV as cv
import cv2
from saveData import *
from numpy import *
#首先完成样本提取工作
print "Get samples..."
dataArr,labelArr = svm.loadImages('trainingDigits')
print "Get samples successfully!"
#开始训练
print "Start training..."
b,alphas = svm.smoP(dataArr,labelArr,200,0.0001,10000,('rbf',10))
storeData(b,'b.txt')
storeData(alphas,'alphas.txt')
print "Training completed!"
#寻找支持向量
dataMat = mat(dataArr)
labelMat = mat(labelArr).transpose()
svInd = nonzero(alphas.A>0)[0]#支持向量索引
sVs = dataMat[svInd]#支持向量
labelSV = labelMat[svInd]#支持向量的标签
storeData(svInd,'svInd.txt')
storeData(sVs,'sVs.txt')
storeData(labelSV,'labelSV.txt')
def test_smo_all():
dataArr, labelArr = svmMLiA.loadDataSet('testSet.txt')
b, alphas = svmMLiA.smoP(dataArr, labelArr, 0.6, 0.001, 40)
print(b)
print(alphas[alphas > 0])
def test_4():
dataArr, labelArr = svmMLiA.loadDataSet('testSet.txt')
b, alphas = svmMLiA.smoP(dataArr, labelArr, 0.6, 0.001, 40)
print shape(alphas[alphas > 0])
for i in range(100):
if alphas[i] > 0.0: print dataArr[i], labelArr[i]
