def applyingSVMToTheMatchmakerDataset():
print '## Applying SVM to the Matchmaker Dataset'
from svm import *
numericalset=advancedclassify.loadnumerical( )
scaledset,scalef=advancedclassify.scaledata(numericalset)
answers,inputs=[r.match for r in scaledset],[r.data for r in scaledset]
inputs1=[]
for i in inputs:
data={};
for m in range(len(i)):
data[m+1]=i[m]
inputs1.append(data)
print inputs1[0]
print 'answers[0]',answers[0],'inputs[0]',inputs[0],'all',len(answers)
''' answers inputs <类别号> <索引1>：<特征值1> <索引2>：<特征值2>.'''
prob = svm_problem(answers,inputs1) #, isKernel=True
param = svm_parameter()
print 'param--->',param,'<----'
m = libsvm.svm_train(prob, param)
newrow=[28.0,-1,-1,26.0,-1,1,2,0.8] # Man doesn't want children, woman does
x0, max_idx = gen_svm_nodearray(scalef(newrow))
print '*** ',libsvm.svm_predict(m,x0)
newrow=[28.0,-1,1,26.0,-1,1,2,0.8] # Both want children
x0, max_idx = gen_svm_nodearray(scalef(newrow))
print '*** ',libsvm.svm_predict(m,x0)
def scalingTheData(): print '## Scaling the Data' reload(advancedclassify) numericalset=advancedclassify.loadnumerical( ) scaledset,scalef=advancedclassify.scaledata(numericalset) avgs=advancedclassify.lineartrain(scaledset) print 'numericalset[0].data',numericalset[0].data print 'numericalset[0].match', numericalset[0].match print advancedclassify.dpclassify(scalef(numericalset[0].data),avgs) print 'numericalset[11].match', numericalset[11].match print advancedclassify.dpclassify(scalef(numericalset[11].data),avgs)
def theKernelTrick1(): print '## The Kernel Trick' numericalset=advancedclassify.loadnumerical( ) scaledset,scalef=advancedclassify.scaledata(numericalset) ssoffset=advancedclassify.getoffset(scaledset) print 'offset',ssoffset print numericalset[0].match print advancedclassify.nlclassify(scalef(numericalset[0].data),scaledset,ssoffset) print numericalset[1].match print advancedclassify.nlclassify(scalef(numericalset[1].data),scaledset,ssoffset) print numericalset[2].match print advancedclassify.nlclassify(scalef(numericalset[2].data),scaledset,ssoffset) newrow=[28.0,-1,-1,26.0,-1,1,2,0.8] # Man doesn't want children, woman does print advancedclassify.nlclassify(scalef(newrow),scaledset,ssoffset) newrow=[28.0,-1,1,26.0,-1,1,2,0.8] # Both want children print advancedclassify.nlclassify(scalef(newrow),scaledset,ssoffset)
def test_pg_197_to_214():
"""
Matchmaker Dataset
"""
import advancedclassify
agesonly = advancedclassify.loadmatch("agesonly.csv", allnum=True)
matchmaker = advancedclassify.loadmatch("matchmaker.csv")
"""
Difficulties with the data
"""
# Scatter plot of mans age vs womans age
# O is a match
# X is not a match
advancedclassify.plotagematches(agesonly)
"""
Basic linear classification
"""
avgs = advancedclassify.lineartrain(agesonly)
print advancedclassify.dpclassify([30,30], avgs)
print advancedclassify.dpclassify([30,25], avgs)
print advancedclassify.dpclassify([25,40], avgs)
print advancedclassify.dpclassify([48,20], avgs)
"""
Determing distances using Yahoo! maps
"""
print advancedclassify.milesdistance("cambride, ma", "new york, ny")
"""
Creating the new dataset
"""
numericalset = advancedclassify.loadnumerical()
print numericalset[0].data
"""
Scaling the dataset.
"""
scaledset, scalef = advancedclassify.scaledata(numericalset)
avgs = advancedclassify.lineartrain(scaledset)
print numericalset[0].data
print numericalset[0].match
print advancedclassify.dpclassify(scalef(numericalset[0].data), avgs)
print numericalset[11].match
print advancedclassify.dpclassify(scalef(numericalset[11].data), avgs)
"""
The kernel trick
"""
offset = advancedclassify.getoffset(agesonly)
print offset
print advancedclassify.nlclassify([30, 30], agesonly, offset)
print advancedclassify.nlclassify([30, 25], agesonly, offset)
print advancedclassify.nlclassify([25, 40], agesonly, offset)
# In contrast to linear classification now recognises that
# 48, 20 is not a good match
print advancedclassify.nlclassify([48, 20], agesonly, offset)
ssoffset = advancedclassify.getoffset(scaledset)
# 0
print numericalset[0].match
# 0
print advancedclassify.nlclassify(scalef(numericalset[0].data), scaledset, ssoffset)
# 1
print numericalset[1].match
# 1
print advancedclassify.nlclassify(scalef(numericalset[1].data), scaledset, ssoffset)
# 0
print numericalset[2].match
# 0
# print advancedclassify.nlclassify(scalef(numericalset[2].data), scaledset, ssoffset)
# man doesnt want children, women does, otherwise really gd match
newrow=[28.0, -1, -1, 26.0, -1, 1, 2, 0.8]
# 0
print advancedclassify.nlclassify(scalef(numericalset[0].data), scaledset, ssoffset)
# both want children
newrow=[28.0, -1, 1, 26.0, -1, 1, 2, 0.8]
# 0
print advancedclassify.nlclassify(scalef(numericalset[0].data), scaledset, ssoffset)
label = row.match
if label == 0:
label = False
else:
label = True
actual = arow_model.predict(row.data)
if actual == label:
correct_count += 1
accuracy = correct_count / len(dataset)
print "[%s] correct: %d/%d, accuracy: %f" % ("AROW", correct_count,
len(dataset), accuracy)
# load and scaling data
agesonly = advancedclassify.loadmatch('agesonly.csv', allnum=True)
numericalset = advancedclassify.loadnumerical('matchmaker.csv')
scaledset, scalef = advancedclassify.scaledata(numericalset)
# show agematch distribution
#advancedclassify.plotagematches(agesonly)
#advancedclassify.plotagematches(numericalset)
# agesonly.csv linear classification
classify_and_validate_dp("agesonly", agesonly)
# matchmaker.csv linear classification
classify_and_validate_dp("matchmaker", numericalset)
classify_and_validate_dp("scaled matchmaker", scaledset)
classify_and_validate_arow(scaledset)
agesonly = ad.loadmatch('agesonly.csv', allnum=True)
matchmaker = ad.loadmatch('matchmaker.csv')
# ad.plotagematches(agesonly)
age = []
for line in file('agesonly.csv'):
l = []
for w in line.split(','):
l.append(int(w))
age.append(l)
tree = tr.buildtree(age)
tr.printtree(tree)
tr.drawtree(tree)
print tr.classify(tree, [65, 63])
avgs = ad.lineartrain(agesonly)
print avgs
print ad.dpclassify([30, 25], avgs.values())
print ad.dpclassify([25, 40], avgs.values())
print ad.dpclassify([48, 20], avgs.values())
print tr.classify(tree, [30, 25])
print tr.classify(tree, [25, 40])
print tr.classify(tree, [48, 20])
numericalset = ad.loadnumerical()
numericalset[0].data
def creatingTheNewDataset(): print '## Creating the New Dataset' reload(advancedclassify) numericalset=advancedclassify.loadnumerical( ) print numericalset[0].data
# print('<----load data---->')
# agesonly = advancedclassify.loadmatch('agesonly.csv', allnum=True)
# matchmaker = advancedclassify.loadmatch('matchmaker.csv')
# advancedclassify.plotagematches(agesonly)
#
#
# # scikit svm
# print('<----Use Scikit Svm---->')
# X = [[1, 0, 1], [-1, 0, -1]]
# y = [1, -1]
# clf = svm.SVC()
# print(clf.fit(X, y))
# print(clf.predict([[1, 1, 1]]))
# Applying SVM to the Matchmaker Dataset
numericalset = advancedclassify.loadnumerical()
print(numericalset[0].data)
scaledset, scalef = advancedclassify.scaledata(numericalset)
answers, input = [r.match for r in scaledset], [r.data for r in scaledset]
clf = svm.SVC()
print(clf.fit(input, answers))
print(clf.predict([scalef([28, -1, 1, 26, -1, 1, 2, 0])]))
# a sample of cross_validating
print('<----A sample of cross_validating---->')
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(input,
answers,
test_size=0.7,
random_state=0)
def get_numericalset():
numericalset = advancedclassify.loadnumerical()
print ">>> numericalset[0].data"
print numericalset[0].data
return numericalset
import advancedclassify
agesonly = advancedclassify.loadmatch('agesonly.csv', allnum = True)
matchmaker = advancedclassify.loadmatch('matchmaker.csv')
reload(advancedclassify)
avgs = advancedclassify.lineartrain(agesonly)
reload(advancedclassify)
advancedclassify.dpclassify([30,30], avgs)
advancedclassify.dpclassify([48,32], avgs)
advancedclassify.dpclassify([25,40], avgs)
advancedclassify.dpclassify([48,20], avgs)
reload(advancedclassify)
numericalset = advancedclassify.loadnumerical()
numericalset[0].data
reload(advancedclassify)
scaledset, scalef = advancedclassify.scaledata(numericalset)
avgs = advancedclassify.lineartrain(scaledset)
numericalset[0].data
reload(advancedclassify)
offset = advancedclassify.getoffset(agesonly)
advancedclassify.nlclassify([30,30], agesonly, offset)
advancedclassify.nlclassify([48,32], agesonly, offset)
advancedclassify.nlclassify([25,40], agesonly, offset)
advancedclassify.nlclassify([48,20], agesonly, offset)
def test_pg_197_to_214():
"""
Matchmaker Dataset
"""
import advancedclassify
agesonly = advancedclassify.loadmatch("agesonly.csv", allnum=True)
matchmaker = advancedclassify.loadmatch("matchmaker.csv")
"""
Difficulties with the data
"""
# Scatter plot of mans age vs womans age
# O is a match
# X is not a match
advancedclassify.plotagematches(agesonly)
"""
Basic linear classification
"""
avgs = advancedclassify.lineartrain(agesonly)
print advancedclassify.dpclassify([30, 30], avgs)
print advancedclassify.dpclassify([30, 25], avgs)
print advancedclassify.dpclassify([25, 40], avgs)
print advancedclassify.dpclassify([48, 20], avgs)
"""
Determing distances using Yahoo! maps
"""
print advancedclassify.milesdistance("cambride, ma", "new york, ny")
"""
Creating the new dataset
"""
numericalset = advancedclassify.loadnumerical()
print numericalset[0].data
"""
Scaling the dataset.
"""
scaledset, scalef = advancedclassify.scaledata(numericalset)
avgs = advancedclassify.lineartrain(scaledset)
print numericalset[0].data
print numericalset[0].match
print advancedclassify.dpclassify(scalef(numericalset[0].data), avgs)
print numericalset[11].match
print advancedclassify.dpclassify(scalef(numericalset[11].data), avgs)
"""
The kernel trick
"""
offset = advancedclassify.getoffset(agesonly)
print offset
print advancedclassify.nlclassify([30, 30], agesonly, offset)
print advancedclassify.nlclassify([30, 25], agesonly, offset)
print advancedclassify.nlclassify([25, 40], agesonly, offset)
# In contrast to linear classification now recognises that
# 48, 20 is not a good match
print advancedclassify.nlclassify([48, 20], agesonly, offset)
ssoffset = advancedclassify.getoffset(scaledset)
# 0
print numericalset[0].match
# 0
print advancedclassify.nlclassify(scalef(numericalset[0].data), scaledset,
ssoffset)
# 1
print numericalset[1].match
# 1
print advancedclassify.nlclassify(scalef(numericalset[1].data), scaledset,
ssoffset)
# 0
print numericalset[2].match
# 0
# print advancedclassify.nlclassify(scalef(numericalset[2].data), scaledset, ssoffset)
# man doesnt want children, women does, otherwise really gd match
newrow = [28.0, -1, -1, 26.0, -1, 1, 2, 0.8]
# 0
print advancedclassify.nlclassify(scalef(numericalset[0].data), scaledset,
ssoffset)
# both want children
newrow = [28.0, -1, 1, 26.0, -1, 1, 2, 0.8]
# 0
print advancedclassify.nlclassify(scalef(numericalset[0].data), scaledset,
ssoffset)
