def make_use_w2v_big(self):
x_all = self.__vectors_provider.get_w2v_big_vectors()
y_all = self.__data_source.get_y()
# TODO here grid search
estimators = []
part1 = LogisticRegression(C=0.5, solver='liblinear')
estimators.append(('logistic', part1))
part2 = SVC(C=10, kernel='rbf')
estimators.append(('svc', part2))
part3 = KNeighborsClassifier(algorithm='auto',
metric='minkowski',
weights='distance')
estimators.append(('knn', part3))
model1 = VotingClassifier(estimators)
cross_val_accuracy, cross_val_f1, train_accuracy, train_f1, test_accuracy, test_f1, y_true, y_pred \
= Evaluator.evaluate(model1, x_all, y_all)
self.__visualizer.show_results(self.__CLASSIFIER_NAME, "model1",
"Word2VecBig", cross_val_accuracy,
cross_val_f1, train_accuracy, train_f1,
test_accuracy, test_f1, y_true, y_pred)
def make_use_w2v_with_tfidf(self):
x_all = self.__vectors_provider.get_w2v_tfidf_vectors()
y_all = self.__data_source.get_y()
# TODO here grid search
model1 = SVC(C=1, kernel='linear', probability=True)
cross_val_accuracy, cross_val_f1, train_accuracy, train_f1, test_accuracy, test_f1, y_true, y_pred \
= Evaluator.evaluate(model1, x_all, y_all)
self.__visualizer.show_results(
self.__CLASSIFIER_NAME, "(C=1, kernel='linear', probability=True)",
"Word2Vec&TF-IDF", cross_val_accuracy, cross_val_f1,
train_accuracy, train_f1, test_accuracy, test_f1, y_true, y_pred)
def make_use_w2v_with_tfidf(self):
x_all = self.__vectors_provider.get_w2v_tfidf_vectors()
y_all = self.__data_source.get_y()
# TODO here grid search
model1 = LogisticRegression(C=1.0, solver='sag', n_jobs=-1)
cross_val_accuracy, cross_val_f1, train_accuracy, train_f1, test_accuracy, test_f1, y_true, y_pred \
= Evaluator.evaluate(model1, x_all, y_all)
self.__visualizer.show_results(self.__CLASSIFIER_NAME,
"(C=1.0, solver='sag')",
"Word2Vec&TF-IDF", cross_val_accuracy,
cross_val_f1, train_accuracy, train_f1,
test_accuracy, test_f1, y_true, y_pred)
def make_use_w2v_with_tfidf(self):
x_all = self.__vectors_provider.get_w2v_tfidf_vectors()
y_all = self.__data_source.get_y()
# TODO here grid search
model1 = KNeighborsClassifier(algorithm='auto', metric='minkowski', weights='distance')
cross_val_accuracy, cross_val_f1, train_accuracy, train_f1, test_accuracy, test_f1, y_true, y_pred \
= Evaluator.evaluate(model1, x_all, y_all)
self.__visualizer.show_results(self.__CLASSIFIER_NAME, "(algorithm='auto', metric='minkowski', weights='distance')",
"Word2Vec&TF-IDF",
cross_val_accuracy, cross_val_f1,
train_accuracy, train_f1,
test_accuracy, test_f1, y_true, y_pred)
def make_use_w2v_old(self):
""" Обучает старый датасет с использованием w2v обученного на новом """
x_all = self.__vectors_provider.get_w2v_old_vectors()
y_all = self.__data_source.get_y()
# TODO here grid search
model1 = SVC(C=1, kernel='linear', probability=True)
cross_val_accuracy, cross_val_f1, train_accuracy, train_f1, test_accuracy, test_f1, y_true, y_pred \
= Evaluator.evaluate(model1, x_all, y_all)
self.__visualizer.show_results(
self.__CLASSIFIER_NAME, "(C=1, kernel='linear', probability=True)",
"Word2VecNewOld", cross_val_accuracy, cross_val_f1, train_accuracy,
train_f1, test_accuracy, test_f1, y_true, y_pred)
def make_use_w2v_old(self):
""" Обучает старый датасет с использованием w2v обученного на новом """
x_all = self.__vectors_provider.get_w2v_old_vectors()
y_all = self.__data_source.get_y()
# TODO here grid search
model1 = LogisticRegression(C=1.0, solver='sag', n_jobs=-1)
cross_val_accuracy, cross_val_f1, train_accuracy, train_f1, test_accuracy, test_f1, y_true, y_pred \
= Evaluator.evaluate(model1, x_all, y_all)
self.__visualizer.show_results(self.__CLASSIFIER_NAME,
"(C=1.0, solver='sag')",
"Word2VecNewOld", cross_val_accuracy,
cross_val_f1, train_accuracy, train_f1,
test_accuracy, test_f1, y_true, y_pred)
def make_use_w2v_old(self):
""" Обучает старый датасет с использованием w2v обученного на новом """
x_all = self.__vectors_provider.get_w2v_old_vectors()
y_all = self.__data_source.get_y()
# TODO here grid search
model1 = KNeighborsClassifier(algorithm='auto', metric='minkowski', weights='distance')
cross_val_accuracy, cross_val_f1, train_accuracy, train_f1, test_accuracy, test_f1, y_true, y_pred \
= Evaluator.evaluate(model1, x_all, y_all)
self.__visualizer.show_results(self.__CLASSIFIER_NAME, "(algorithm='auto', metric='minkowski', weights='distance')",
"Word2VecNewOld",
cross_val_accuracy, cross_val_f1,
train_accuracy, train_f1,
test_accuracy, test_f1, y_true, y_pred)
def make_use_w2v(self):
x_all = self.__vectors_provider.get_w2v_vectors()
y_all = self.__data_source.get_y()
# TODO here grid search
model1 = LogisticRegression(C=1.0, solver='sag', n_jobs=-1)
Evaluator.cross_probabilities(model=model1,
x_all=x_all,
y_all=y_all,
data_source=self.__data_source,
method='w2v')
cross_val_accuracy, cross_val_f1, train_accuracy, train_f1, test_accuracy, test_f1, y_true, y_pred \
= Evaluator.evaluate(model1, x_all, y_all)
self.__visualizer.show_results(self.__CLASSIFIER_NAME,
"(C=1.0, solver='sag')", "Word2Vec",
cross_val_accuracy, cross_val_f1,
train_accuracy, train_f1, test_accuracy,
test_f1, y_true, y_pred)