# print "%s score : %f" % (m_classifiers_name_it, m_cross_validation_score[m_classifiers_name_it])
print "training classifiers:"
TraininAllClassifiers(m_classifiers, train_data, train_label, train_data,
train_label)
save_classifiers(m_classifiers)
###########################################
###### preditct and show the result #######
###########################################
try:
notexist
predict_res
except NameError:
print "testing in real test data:"
predict_res = PrediectinAllClassifiers(m_classifiers, test_data)
else:
print "Predict has been extracted!"
m_predictMode = ModethePredict(test_data, predict_res)
PlotPredictRes(m_predictMode, 'MODE')
df = pd.DataFrame(m_predictMode)
df = df.T
df.to_csv("../prefile.txt", sep=' ', index=False, header=False)
# save center and scalar to file
m_cent = m_normalizer.robust_scaler.center_
m_scal = m_normalizer.robust_scaler.scale_
m_cent = pd.DataFrame(m_cent)
m_scal = pd.DataFrame(m_scal)
m_cent.to_csv("../center.txt", sep=' ', index=False, header=False)
train_data, test_data, train_label, test_label = cross_validation.train_test_split(
train_data, train_label, test_size=0.1)
print "training classifiers:"
TraininAllClassifiers(train_data, train_label, test_data, test_label)
###########################################
###### preditct and show the result #######
###########################################
test_naivedata = featureOf_TestinTrain
try:
notexist
predictRes
except NameError:
print "testing in train data:"
predictRes = PrediectinAllClassifiers(test_naivedata)
else:
print "Predict has been extracted!"
m_predictMode = ModethePredict(test_naivedata, predictRes)
PlotTestSeqandPredictRes(
m_normalized_traindata.loc[121200:149000]['accelerometerX'], m_predictMode,
'MODE')
a = m_predictMode
test_realdata = featureOf_TestinReal
try:
notexist
predictResinReal
except NameError:
print "testing in real test data:"
predictResinReal = PrediectinAllClassifiers(test_realdata)
else:
train_data = train_data.append(featureOf_Train[i].loc[j])
train_label.append(label_numlabel[label_name[i]])
###########################################
############# normalize data ##############
###########################################
print "normalize feature:"
m_normalizer = processingFeature.Normalizer("robust", train_data)
train_data = m_normalizer.normalizer(train_data)
featureOf_TestinReal = m_normalizer.normalizer(featureOf_TestinReal)
print "PCA feature:"
m_pcaor = processingFeature.PCAor("normal", train_data, 50)
train_data = m_pcaor.pcaor(train_data)
featureOf_TestinReal = m_pcaor.pcaor(featureOf_TestinReal)
train_data, test_data, train_label, test_label = cross_validation.train_test_split(
train_data, train_label, test_size=0.1)
m_classifiers = load_classifiers()
###########################################
###### preditct and show the result #######
###########################################
test_realdata = featureOf_TestinReal
predictResinReal = PrediectinAllClassifiers(m_classifiers, test_realdata)
m_predictMode = ModethePredict(test_realdata, predictResinReal)
PlotTestSeqandPredictRes(m_normalized_testdata['accelerometerX'],
m_predictMode, 'MODE')
df = pd.DataFrame(m_predictMode)
df = df.T
df.to_csv("../prefile", index=False, header=False)
train_data, test_data, train_label, test_label = cross_validation.train_test_split(
train_data, train_label, test_size=0.5)
TraininAllClassifiers(train_data, train_label, test_data, test_label)
###########################################
###### preditct and show the result #######
###########################################
# no need to reshape test data and predict
test_data = featureOfTest
try:
#notexist
predictRes
except NameError:
predictRes = PrediectinAllClassifiers(test_data)
else:
print "Predict has been extracted!"
# get the mode of different classifiers, present a vote function
def ModethePredict():
from scipy.stats import mode
predictMode = []
for i in range(len(test_data)):
predictMode.append(
mode([ #predictRes['LR'][i],
predictRes['KNN']
[i], #predictRes['KNN'][i],predictRes['KNN'][i],
predictRes['RF'][i],
predictRes['GBDT'][i]
for i in range(len(test_data)):
predictMode.append(
mode([ #predictRes['LR'][i],
predictRes['KNN']
[i], #predictRes['KNN'][i],predictRes['KNN'][i],
predictRes['RF'][i],
predictRes['GBDT'][i] #,predictRes['GBDT'][i]
])[0][0])
return predictMode
try:
notexist
predictRes
except NameError:
predictRes = PrediectinAllClassifiers(test_naivedata)
else:
print "Predict has been extracted!"
m_predictMode = ModethePredict(test_naivedata, predictRes)
PlotTestSeqandPredictRes(
m_normalized_traindata.loc[121200:149000]['accelerometerX'], m_predictMode,
'MODE')
try:
notexist
predictResinReal
except NameError:
print "testing in real test data:"
predictResinReal = PrediectinAllClassifiers(test_realdata)
else:
print "Predict has been extracted!"