def get_the_iter_accucy(max_true_vale=70, model="model/GNB.model", iter_times=3):
try:
if iter_times > 0:
print "left " + str(iter_times) + " iteration"
a, b = load_data_X_Y('train')
X, y = shuffle_two_list_X_Y(a, b)
gnb = GaussianNB()
gnb.fit(X, y)
testX, testY = load_data_X_Y('test')
true_ans = 0
for itemX in range(len(testX)):
predict_result = gnb.predict(testX[itemX])
vote_vale = vote_the_max_times(predict_result)
if vote_vale == testY[itemX]:
true_ans += 1
true_ans_percent = true_ans * 100 / len(testX)
print true_ans_percent
if true_ans_percent > max_true_vale:
print "good job, the new record:"
joblib.dump(gnb, model)
print "update the model"
else:
true_ans_percent = max_true_vale
get_the_iter_accucy(true_ans_percent, model, iter_times - 1)
except ValueError:
pass
return 0
def get_accucy(model="model/GB/GB.model"):
try:
clf = joblib.load(model)
print 'load GB model successfully'
except IOError:
print 'GB model file doesnt exist, Train first'
a, b = load_data_X_Y('train')
# X, y = shuffle_two_list_X_Y(a, b)
X, y = a, b
#clf = gaussian_process.GaussianProcess(theta0=1e-2, thetaL=1e-4, thetaU=1e-1)
clf = GradientBoostingClassifier(n_estimators=100,
learning_rate=1.0,
max_depth=1,
random_state=0)
clf.fit(X, y)
joblib.dump(clf, "model/GB/GB.model")
print "update the model"
testX, testY = load_data_X_Y('test')
true_ans = 0
for itemX in range(len(testX)):
predict_result = clf.predict(testX[itemX])
vote_vale = vote_the_max_times(predict_result)
if vote_vale == testY[itemX]:
true_ans += 1
true_ans_percent = true_ans * 100 / len(testX)
return true_ans_percent
# get_the_iter_accucy()
# print get_accucy()
def get_accucy(model="model/GNB.model"):
try:
gnb = joblib.load(model)
print 'load GNBClassifier successfully'
except IOError, ValueError:
print 'GNB classifer file doesnt exist, Train first'
a, b = load_data_X_Y('train')
X, y = shuffle_two_list_X_Y(a, b)
gnb = GaussianNB()
gnb.fit(X, y)
def get_accucy(model="model/DT.model"):
try:
dt = joblib.load(model)
print 'load DTClassifier successfully'
except IOError, ValueError:
print 'DT classifer file doesnt exist, Train first'
a, b = load_data_X_Y('train')
X, y = shuffle_two_list_X_Y(a, b)
dt = DecisionTreeClassifier()
dt.fit(X, y)
def get_accucy(model="model/SGD.model"):
try:
sgd = joblib.load(model)
print 'load SGDClassifier successfully'
except IOError:
print 'SGD classifer file doesnt exist, Train first'
a, b = load_data_X_Y('train')
X, y = shuffle_two_list_X_Y(a, b)
sgd = SGDClassifier(loss="hinge", penalty="l2", n_iter=10000)
sgd.fit(X, y)
testX, testY = load_data_X_Y('test')
true_ans = 0
for itemX in range(len(testX)):
predict_result = sgd.predict(testX[itemX])
vote_vale = vote_the_max_times(predict_result)
if vote_vale == testY[itemX]:
true_ans += 1
true_ans_percent = true_ans * 100 / len(testX)
return true_ans_percent
def get_the_iter_accucy(max_true_vale=76,
model="model/GB/GB.model",
iter_times=1):
try:
if iter_times > 0:
print "left " + str(iter_times) + " iteration"
a, b = load_data_X_Y('train')
# X, y = shuffle_two_list_X_Y(a, b)
X, y = a, b
#clf = gaussian_process.GaussianProcess(theta0=1e-2, thetaL=1e-4, thetaU=1e-1)
clf = GradientBoostingClassifier(n_estimators=100,
learning_rate=1.0,
max_depth=1,
random_state=0)
clf.fit(X, y)
joblib.dump(clf, "model/GB/GB.model")
print "update the model"
testX, testY = load_data_X_Y('test')
true_ans = 0
for itemX in range(len(testX)):
predict_result = clf.predict(testX[itemX])
vote_vale = vote_the_max_times(predict_result)
if vote_vale == testY[itemX]:
true_ans += 1
true_ans_percent = true_ans * 100 / len(testX)
print true_ans_percent
if true_ans_percent > max_true_vale:
print "good job, the new record:"
joblib.dump(clf, model)
print "update the model"
else:
true_ans_percent = max_true_vale
get_the_iter_accucy(true_ans_percent, model, iter_times - 1)
except ValueError:
pass
return 0
def start_calssification_SGD(wav):
try:
sgd = joblib.load('model/SGD.model')
print 'load SGDClassifier successfully'
except IOError:
print 'SGD classifer file doesnt exist, Train first'
a, b = load_data_X_Y('train')
X, y = shuffle_two_list_X_Y(a, b)
sgd = SGDClassifier(loss="hinge", penalty="l2", n_iter=10000)
sgd.fit(X, y)
predict_result = sgd.predict(load_data_user_chose(wav))
vote_vale = vote_the_max_times(predict_result)
speaker_name = get_speaker_name(vote_vale)
return speaker_name
def start_calssification_GNB(wav):
try:
gnb = joblib.load('model/GNB.model')
print 'load GNBClassifier successfully'
except IOError:
print 'GNB classifer file doesnt exist, Train first'
a, b = load_data_X_Y('train')
X, y = shuffle_two_list_X_Y(a, b)
gnb = GaussianNB()
gnb.fit(X, y)
joblib.dump(gnb, 'model/GNB.model')
print "update the model"
predict_result = gnb.predict(load_data_user_chose(wav))
vote_vale = vote_the_max_times(predict_result)
speaker_name = get_speaker_name(vote_vale)
return speaker_name
def start_calssification_DT(wav):
try:
dt = joblib.load('model/DT.model')
print 'load DTClassifier successfully'
except IOError:
print 'DT classifer file doesnt exist, Train first'
a, b = load_data_X_Y('train')
X, y = shuffle_two_list_X_Y(a, b)
dt = DecisionTreeClassifier()
dt.fit(X, y)
joblib.dump(dt, 'model/DT.model')
print "update the model"
predict_result = dt.predict(load_data_user_chose(wav))
vote_vale = vote_the_max_times(predict_result)
speaker_name = get_speaker_name(vote_vale)
return speaker_name
def start_GradientBoostingClassifier(wav):
try:
clf = joblib.load('model/GB/GB.model')
print 'load GB model successfully'
except IOError:
print 'GB model file doesnt exist, Train first'
a, b = load_data_X_Y('train')
# X, y = shuffle_two_list_X_Y(a, b)
X, y = a, b
#clf = gaussian_process.GaussianProcess(theta0=1e-2, thetaL=1e-4, thetaU=1e-1)
clf = GradientBoostingClassifier(n_estimators=100,
learning_rate=1.0,
max_depth=1,
random_state=0)
clf.fit(X, y)
joblib.dump(clf, "model/GB/GB.model")
print "update the model"
predict_result = clf.predict(load_data_user_chose(wav))
vote_vale = vote_the_max_times(predict_result)
speaker_name = get_speaker_name(vote_vale)
return speaker_name
pass
return 0
def get_accucy(model="model/GNB.model"):
try:
gnb = joblib.load(model)
print 'load GNBClassifier successfully'
except IOError, ValueError:
print 'GNB classifer file doesnt exist, Train first'
a, b = load_data_X_Y('train')
X, y = shuffle_two_list_X_Y(a, b)
gnb = GaussianNB()
gnb.fit(X, y)
testX, testY = load_data_X_Y('test')
true_ans = 0
for itemX in range(len(testX)):
predict_result = gnb.predict(testX[itemX])
vote_vale = vote_the_max_times(predict_result)
if vote_vale == testY[itemX]:
true_ans += 1
true_ans_percent = true_ans * 100 / len(testX)
return true_ans_percent
# get_the_iter_accucy()
# print get_accucy()
#clf = gaussian_process.GaussianProcess(theta0=1e-2, thetaL=1e-4, thetaU=1e-1)
clf = GradientBoostingClassifier(n_estimators=100,
learning_rate=1.0,
max_depth=1,
random_state=0)
clf.fit(X, y)
joblib.dump(clf, "model/GB/GB.model")
print "update the model"
predict_result = clf.predict(load_data_user_chose(wav))
vote_vale = vote_the_max_times(predict_result)
speaker_name = get_speaker_name(vote_vale)
return speaker_name
test1, test2 = load_data_X_Y('train')
# print start_calssification_SGD()
def get_the_iter_accucy(max_true_vale=76,
model="model/GB/GB.model",
iter_times=1):
try:
if iter_times > 0:
print "left " + str(iter_times) + " iteration"
a, b = load_data_X_Y('train')
# X, y = shuffle_two_list_X_Y(a, b)
X, y = a, b
#clf = gaussian_process.GaussianProcess(theta0=1e-2, thetaL=1e-4, thetaU=1e-1)
clf = GradientBoostingClassifier(n_estimators=100,
learning_rate=1.0,