def voting_classify(X_train, y_train, X_cv, y_cv, X_test, y_test):
from sklearn.ensemble import GradientBoostingClassifier, VotingClassifier, RandomForestClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.naive_bayes import GaussianNB
clf1 = GradientBoostingClassifier(n_estimators=200)
clf2 = RandomForestClassifier(random_state=0, n_estimators=500)
clf3 = LogisticRegression(random_state=1)
clf4 = GaussianNB()
clf = VotingClassifier(
estimators=[
('gbdt', clf1),
('rf', clf2),
('lr', clf3),
('nb', clf4),
# ('xgboost',clf5),
],
voting='soft')
clf.fit(X_train, y_train)
pre_y_train = clf.predict(X_train)
pre_y_cv = clf.predict(X_cv)
pre_y_test = clf.predict(X_test)
print("voting_classify train Metrics : {0}".format(
PRF(y_train, pre_y_train)))
print("voting_classify cv Metrics : {0}".format(PRF(y_cv, pre_y_cv)))
print("voting_classify test Metrics : {0}".format(
PRF(y_test, pre_y_test)))
print("Test PRF : {0}".format(
precision_recall_fscore_support(y_test, pre_y_test)))
print('The Accuracy of ' + 'Voting' + ' is :', clf.score(X_test, y_test))
print(classification_report(y_test, pre_y_test))
return clf
def svm_classify(X_train, y_train, X_cv, y_cv, X_test, y_test):
from sklearn import svm
clf = svm.SVC(kernel='linear', C=1e4, gamma=0.01)
clf.fit(X_train, y_train)
pre_y_train = clf.predict(X_train)
pre_y_cv = clf.predict(X_cv)
pre_y_test = clf.predict(X_test)
print("svm train Metrics : {0}".format(PRF(y_train, pre_y_train)))
print("svm cv Metrics : {0}".format(PRF(y_cv, pre_y_cv)))
print("svm test Metrics : {0}".format(PRF(y_test, pre_y_test)))
print("Test PRF : {0}".format(
precision_recall_fscore_support(y_test, pre_y_test)))
print('The Accuracy of ' + 'SVC' + ' is :', clf.score(X_test, y_test))
print(classification_report(y_test, pre_y_test))
return clf
def xgboost_classify(X_train, y_train, X_cv, y_cv, X_test, y_test):
import xgboost
clf = xgboost.XGBClassifier()
clf.fit(X_train, y_train)
pre_y_train = clf.predict(X_train)
pre_y_cv = clf.predict(X_cv)
pre_y_test = clf.predict(X_test)
print("xgboost train Metrics : {0}".format(PRF(y_train, pre_y_train)))
print("xgboost cv Metrics : {0}".format(PRF(y_cv, pre_y_cv)))
print("xgboost test Metrics : {0}".format(PRF(y_test, pre_y_test)))
print("Test PRF : {0}".format(
precision_recall_fscore_support(y_test, pre_y_test)))
print('The Accuracy of ' + 'xgboost' + ' is :', clf.score(X_test, y_test))
print(classification_report(y_test, pre_y_test))
return clf
def da_classify(X_train, y_train, X_cv, y_cv, X_test, y_test):
from sklearn.discriminant_analysis import QuadraticDiscriminantAnalysis
clf = QuadraticDiscriminantAnalysis()
clf.fit(X_train, y_train)
pre_y_train = clf.predict(X_train)
pre_y_cv = clf.predict(X_cv)
pre_y_test = clf.predict(X_test)
print("da train Metrics : {0}".format(PRF(y_train, pre_y_train)))
print("da cv Metrics : {0}".format(PRF(y_cv, pre_y_cv)))
print("da test Metrics : {0}".format(PRF(y_test, pre_y_test)))
print("Test PRF : {0}".format(
precision_recall_fscore_support(y_test, pre_y_test)))
print('The Accuracy of ' + 'da' + ' is :', clf.score(X_test, y_test))
print(classification_report(y_test, pre_y_test))
return clf
def nb_classify(X_train, y_train, X_cv, y_cv, X_test, y_test):
from sklearn.naive_bayes import GaussianNB
clf = GaussianNB()
clf.fit(X_train, y_train)
pre_y_train = clf.predict(X_train)
pre_y_cv = clf.predict(X_cv)
pre_y_test = clf.predict(X_test)
print("nb train Metrics : {0}".format(PRF(y_train, pre_y_train)))
print("nb cv Metrics : {0}".format(PRF(y_cv, pre_y_cv)))
print("nb test Metrics : {0}".format(PRF(y_test, pre_y_test)))
print("Test PRF : {0}".format(
precision_recall_fscore_support(y_test, pre_y_test)))
print('The Accuracy of ' + 'nb' + ' is :', clf.score(X_test, y_test))
print(classification_report(y_test, pre_y_test))
return clf
def knn_classify(X_train, y_train, X_cv, y_cv, X_test, y_test):
from sklearn.neighbors import KNeighborsClassifier
clf = KNeighborsClassifier(n_neighbors=5)
clf.fit(X_train, y_train)
pre_y_train = clf.predict(X_train)
pre_y_cv = clf.predict(X_cv)
pre_y_test = clf.predict(X_test)
print("knn train Metrics : {0}".format(PRF(y_train, pre_y_train)))
print("knn cv Metrics : {0}".format(PRF(y_cv, pre_y_cv)))
print("knn test Metrics : {0}".format(PRF(y_test, pre_y_test)))
print("Test PRF : {0}".format(
precision_recall_fscore_support(y_test, pre_y_test)))
print('The Accuracy of ' + 'KNN' + ' is :', clf.score(X_test, y_test))
print(classification_report(y_test, pre_y_test))
return clf
def decisionTree_classify(X_train, y_train, X_cv, y_cv, X_test, y_test):
from sklearn.tree import DecisionTreeClassifier
clf = DecisionTreeClassifier(max_depth=3)
clf.fit(X_train, y_train)
pre_y_train = clf.predict(X_train)
pre_y_cv = clf.predict(X_cv)
pre_y_test = clf.predict(X_test)
print("DT train Metrics : {0}".format(PRF(y_train, pre_y_train)))
print("DT cv Metrics : {0}".format(PRF(y_cv, pre_y_cv)))
print("DT test Metrics : {0}".format(PRF(y_test, pre_y_test)))
print("Test PRF : {0}".format(
precision_recall_fscore_support(y_test, pre_y_test)))
print('The Accuracy of ' + 'decisionTree' + ' is :',
clf.score(X_test, y_test))
print(classification_report(y_test, pre_y_test))
return clf
def lr_classify(X_train, y_train, X_cv, y_cv, X_test, y_test):
from sklearn.linear_model import LogisticRegression
clf = LogisticRegression(C=1e5)
clf.fit(X_train, y_train)
pre_y_train = clf.predict(X_train)
pre_y_cv = clf.predict(X_cv)
pre_y_test = clf.predict(X_test)
print("lr train Metrics : {0}".format(PRF(y_train, pre_y_train)))
print("lr cv Metrics : {0}".format(PRF(y_cv, pre_y_cv)))
print("lr test Metrics : {0}".format(PRF(y_test, pre_y_test)))
print("Test PRF : {0}".format(
precision_recall_fscore_support(y_test, pre_y_test)))
print('The Accuracy of ' + 'bagging_knn' + ' is :',
clf.score(X_test, y_test))
print(classification_report(y_test, pre_y_test))
return clf
def GBDT_classify(X_train, y_train, X_cv, y_cv, X_test, y_test):
from sklearn.ensemble import GradientBoostingClassifier
clf = GradientBoostingClassifier(n_estimators=200)
clf.fit(X_train, y_train)
pre_y_train = clf.predict(X_train)
pre_y_cv = clf.predict(X_cv)
pre_y_test = clf.predict(X_test)
print("GBDT_classify train Metrics : {0}".format(
PRF(y_train, pre_y_train)))
print("GBDT_classify cv Metrics : {0}".format(PRF(y_cv, pre_y_cv)))
print("GBDT_classify test Metrics : {0}".format(PRF(y_test,
pre_y_test)))
print("Test PRF : {0}".format(
precision_recall_fscore_support(y_test, pre_y_test)))
print('The Accuracy of ' + 'GBDT' + ' is :', clf.score(X_test, y_test))
print(classification_report(y_test, pre_y_test))
return clf
def rf_classify(X_train, y_train, X_cv, y_cv, X_test, y_test):
from sklearn.ensemble import RandomForestClassifier
clf = RandomForestClassifier(random_state=0, n_estimators=100)
clf.fit(X_train, y_train)
pre_y_train = clf.predict(X_train)
pre_y_cv = clf.predict(X_cv)
pre_y_test = clf.predict(X_test)
print("rf train Metrics : {0}".format(PRF(y_train, pre_y_train)))
print("rf cv Metrics : {0}".format(PRF(y_cv, pre_y_cv)))
print("rf test Metrics : {0}".format(PRF(y_test, pre_y_test)))
print("Test PRF : {0}".format(
precision_recall_fscore_support(y_test, pre_y_test)))
print('The Accuracy of ' + 'rf' + ' is :', clf.score(X_test, y_test))
print(classification_report(y_test, pre_y_test))
# fpr, tpr, thresholds = metrics.roc_curve(y_test, pre_y_test)
# print metrics.auc(fpr, tpr)
return clf
def MLP_classify(X_train, y_train, X_cv, y_cv, X_test, y_test):
from sklearn.neural_network import MLPClassifier
from sklearn.neural_network import MLPClassifier
t0 = time()
clf = MLPClassifier(alpha=1)
clf.fit(X_train, y_train)
pre_y_train = clf.predict(X_train)
pre_y_cv = clf.predict(X_cv)
pre_y_test = clf.predict(X_test)
print("voting_classify train Metrics : {0}".format(
PRF(y_train, pre_y_train)))
print("voting_classify cv Metrics : {0}".format(PRF(y_cv, pre_y_cv)))
print("voting_classify test Metrics : {0}".format(
PRF(y_test, pre_y_test)))
print("Test PRF : {0}".format(
precision_recall_fscore_support(y_test, pre_y_test)))
print('The Accuracy of ' + 'MLP' + ' is :', clf.score(X_test, y_test))
print(classification_report(y_test, pre_y_test))
return clf
def bagging_knn_classify(X_train, y_train, X_cv, y_cv, X_test, y_test):
from sklearn.neighbors import KNeighborsClassifier
from sklearn.ensemble import BaggingClassifier
clf = BaggingClassifier(KNeighborsClassifier(),
max_samples=0.5,
max_features=0.5)
clf.fit(X_train, y_train)
pre_y_train = clf.predict(X_train)
pre_y_cv = clf.predict(X_cv)
pre_y_test = clf.predict(X_test)
print("bagging_knn train Metrics : {0}".format(PRF(y_train, pre_y_train)))
print("bagging_knn cv Metrics : {0}".format(PRF(y_cv, pre_y_cv)))
print("bagging_knn test Metrics : {0}".format(PRF(y_test, pre_y_test)))
print("Test PRF : {0}".format(
precision_recall_fscore_support(y_test, pre_y_test)))
print('The Accuracy of ' + 'bagging_knn' + ' is :',
clf.score(X_test, y_test))
print(classification_report(y_test, pre_y_test))
return clf
def cnn_model(X_train, y_train, X_cv, y_cv, X_test, y_test):
'''
Building this model with CNN
'''
# 所有数据的输入形状
max_npara = X_train.shape[2]
max_nsent = X_train.shape[3]
para_conv = 3
sent_conv = 3
embedding_dim = 100
maxlen = 400
# number of convolutional filters to use
nb_filters = 32
# convolution kernel size
nb_conv = 3
# size of pooling area for max pooling
nb_pool = 2
batch_size = 128
nb_classes = 10
nb_epoch = 10
iteraStartEpoch = 50
iterEndEpoch = 81
n_class = 4
# 段落模型
essay_input = Input(shape=(max_npara, max_nsent, embedding_dim))
max_features = 20000
embedding_layer = Embedding(max_features, 128, input_length=maxlen)
x = Convolution2D(nb_filters,
para_conv,
sent_conv,
name='conv1',
border_mode='valid')(embedding_layer)
x = Activation('relu')(x)
x = Convolution2D(nb_filters, para_conv, sent_conv, name='conv2')(x)
x = Activation('relu')(x)
x = MaxPooling2D(pool_size=(nb_pool, nb_pool))(x)
# x = Dropout(0.5)(x)
x = Flatten()(x)
x = Dense(128)(x)
x = Activation('relu')(x)
x = Dropout(0.4)(x)
x = Dense(n_class)(x)
prediction = Activation('softmax')(x)
essay_model = Model(input=essay_input, output=prediction)
essay_model.compile(loss='categorical_crossentropy',
optimizer='adadelta',
metrics=['accuracy'])
# print(essay_model.summary())
# plot(essay_model, to_file='cnn_model.png')
essay_model.fit(X_train,
y_train,
batch_size=16,
epochs=nb_epoch,
validation_data=(X_test, y_test))
pre_y_train = essay_model.predict(X_train)
pre_y_cv = essay_model.predict(X_cv)
pre_y_test = essay_model.predict(X_test)
print("voting_classify train Metrics : {0}".format(
PRF(y_train, pre_y_train)))
print("voting_classify cv Metrics : {0}".format(PRF(y_cv, pre_y_cv)))
print("voting_classify test Metrics : {0}".format(
PRF(y_test, pre_y_test)))
return essay_model