def init_classifier_models(classifier_data_url, test_data_url, fileName):
with open(classifier_data_url, 'rb') as file:
classifier = pickle.load(file)
_service = service()
_test_data = _service.read_csv_data_without_labels(test_data_url)
_val_pred = classifier.predict(_test_data)
_service.save_csv_result(_val_pred, fileName, 'output_csvs')
def parameter_iteration_tunning():
_service = service()
train_features, train_labels = _service.read_csv_data('dataset2/ds2/ds2Train.csv')
validation_features, validation_labels = _service.read_csv_data('dataset2/ds2/ds2Val.csv')
_range = [0.01, 0.001, 0.1, 1, 10]
for index in _range:
_clf = BernoulliNB(alpha=index)
_clf.fit(train_features, train_labels )
pred = _clf.predict(validation_features)
print('BernoulliNB accuracy ' + str(index) + ' is', accuracy_score(validation_labels, pred))
def parameter_iteration_tunning():
_service = service()
train_features, train_labels = _service.read_csv_data(
'dataset1/ds1/ds1Train.csv')
validation_features, validation_labels = _service.read_csv_data(
'dataset1/ds1/ds1Val.csv')
_range = [10, 100, 200, 300, 400, 500, 1000]
for index in _range:
_clf = RandomForestClassifier(n_estimators=index)
_clf.fit(train_features, train_labels)
pred = _clf.predict(validation_features)
print('RandomForest accuracy ' + str(index) + ' is',
accuracy_score(validation_labels, pred))
def load_naive_bayes_bernoulli():
_service = service()
train_features, train_labels = _service.read_csv_data('dataset2/ds2/ds2Train.csv')
validation_features, validation_labels = _service.read_csv_data('dataset2/ds2/ds2Val.csv')
# classifier = BernoulliNB()
classifier = BernoulliNB(alpha=0.001)
classifier = classifier.fit(train_features, train_labels)
validation_predicted = classifier.predict(validation_features)
print(classification_report(validation_labels, validation_predicted))
print(confusion_matrix(validation_labels, validation_predicted))
print("Accuracy Score:", accuracy_score(validation_labels, validation_predicted))
_service.save_csv_result(validation_predicted, 'ds2Val-nb', 'output_csvs')
_service.save_model_to_pkl(classifier, 'ds2Classifier-nb', 'dataset2/models/')
def parameter_iteration_tunning():
_service = service()
train_features, train_labels = _service.read_csv_data(
'dataset1/ds1/ds1Train.csv')
validation_features, validation_labels = _service.read_csv_data(
'dataset1/ds1/ds1Val.csv')
_range = range(1, 30)
for index in _range:
_clf = DecisionTreeClassifier(max_depth=index)
_clf.fit(train_features, train_labels)
pred = _clf.predict(validation_features)
print('Decision Tree accuracy ' + str(index) + ' is',
accuracy_score(validation_labels, pred))
def load_naive_bayes_multinomial():
_service = service()
train_features, train_labels = _service.read_csv_data('dataset1/ds1/ds1Train.csv')
validation_features, validation_labels = _service.read_csv_data('dataset1/ds1/ds1Val.csv')
# classifier = MultinomialNB()
classifier = MultinomialNB(alpha=0.01)
classifier = classifier.fit(train_features, train_labels)
validation_predicted = classifier.predict(validation_features)
print(classification_report(validation_labels, validation_predicted))
print(confusion_matrix(validation_labels, validation_predicted))
print("Accuracy Score:", accuracy_score(validation_labels, validation_predicted))
_service.save_csv_result(validation_predicted, 'nb_multinomial_results', 'dataset1/results/')
_service.save_model_to_pkl(classifier, 'nb_multinomial_results', 'dataset1/models/')
def load_random_forest():
_service = service()
train_features, train_labels = _service.read_csv_data(
'dataset1/ds1/ds1Train.csv')
validation_features, validation_labels = _service.read_csv_data(
'dataset1/ds1/ds1Val.csv')
# classifier = RandomForestClassifier()
classifier = RandomForestClassifier(n_estimators=10000)
classifier = classifier.fit(train_features, train_labels)
validation_predicted = classifier.predict(validation_features)
print(classification_report(validation_labels, validation_predicted))
print(confusion_matrix(validation_labels, validation_predicted))
print("Accuracy Score:",
accuracy_score(validation_labels, validation_predicted))
_service.save_csv_result(validation_predicted, 'ds1Val-3', 'output_csvs')
_service.save_model_to_pkl(classifier, 'ds1Classifier-3',
'dataset1/models/')
def load_decision_tree():
_service = service()
train_features, train_labels = _service.read_csv_data(
'dataset1/ds1/ds1Train.csv')
validation_features, validation_labels = _service.read_csv_data(
'dataset1/ds1/ds1Val.csv')
# classifier = DecisionTreeClassifier()
classifier = DecisionTreeClassifier(criterion='entropy', max_depth=36)
classifier = classifier.fit(train_features, train_labels)
validation_predicted = classifier.predict(validation_features)
print(classification_report(validation_labels, validation_predicted))
print(confusion_matrix(validation_labels, validation_predicted))
print("Accuracy Score:",
accuracy_score(validation_labels, validation_predicted))
_service.save_csv_result(validation_predicted, 'ds1Val-dt', 'output_csvs')
_service.save_model_to_pkl(classifier, 'ds1Classifier-dt',
'dataset1/models/')
def load_kneighbors():
_service = service()
train_features, train_labels = _service.read_csv_data(
'dataset1/ds1/ds1Train.csv')
validation_features, validation_labels = _service.read_csv_data(
'dataset1/ds1/ds1Val.csv')
# classifier = KNeighborsClassifier(n_neighbors=1)
classifier = KNeighborsClassifier(n_neighbors=1)
classifier = classifier.fit(train_features, train_labels)
validation_predicted = classifier.predict(validation_features)
print(classification_report(validation_labels, validation_predicted))
print(confusion_matrix(validation_labels, validation_predicted))
print("Accuracy Score:",
accuracy_score(validation_labels, validation_predicted))
_service.save_csv_result(validation_predicted, 'k_neighbors_results',
'dataset1/results/')
_service.save_model_to_pkl(classifier, 'k_neighbors_results',
'dataset1/models/')
def load_naive_bayes_complement():
_service = service()
train_features, train_labels = _service.read_csv_data(
'dataset2/ds2/ds2Train.csv')
validation_features, validation_labels = _service.read_csv_data(
'dataset2/ds2/ds2Val.csv')
# classifier = ComplementNB()
classifier = ComplementNB(alpha=1000)
classifier = classifier.fit(train_features, train_labels)
validation_predicted = classifier.predict(validation_features)
print(classification_report(validation_labels, validation_predicted))
print(confusion_matrix(validation_labels, validation_predicted))
print("Accuracy Score:",
accuracy_score(validation_labels, validation_predicted))
_service.save_csv_result(validation_predicted, 'nb_complement_results',
'dataset2/results/')
_service.save_model_to_pkl(classifier, 'nb_complement_results',
'dataset2/models/')
