def perceptron(training, test, output):
training_data = np.loadtxt(training, delimiter=',')
test_data = np.loadtxt(test, delimiter=',')
per = Perceptron()
per.fit(training_data[:, :-1], training_data[:, -1])
test_prediction = per.predict(test_data[:, :-1])
helper.output_computed_metrics(test_data, test_prediction, output)
def base_decision_tree(training, test, output):
training_data = np.loadtxt(training, delimiter=',')
test_data = np.loadtxt(test, delimiter=',')
dt = DecisionTreeClassifier(criterion='entropy')
dt.fit(training_data[:, :-1], training_data[:, -1])
test_prediction = dt.predict(test_data[:, :-1])
helper.output_computed_metrics(test_data, test_prediction, output)
def gaussian_naive_bayes(training, test, output):
training_data = np.loadtxt(training, delimiter=',')
test_data = np.loadtxt(test, delimiter=',')
gnb = GaussianNB()
gnb.fit(training_data[:, :-1], training_data[:, -1])
test_prediction = gnb.predict(test_data[:, :-1]).astype(int)
helper.output_computed_metrics(test_data, test_prediction, output)
def base_mlp(training, test, output):
training_data = np.loadtxt(training, delimiter=',')
test_data = np.loadtxt(test, delimiter=',')
mlp = MLPClassifier(hidden_layer_sizes=(1, 100), activation='logistic', solver='sgd')
mlp.fit(training_data[:, :-1], training_data[:, -1])
test_prediction = mlp.predict(test_data[:, :-1])
helper.output_computed_metrics(test_data, test_prediction, output)
def best_decision_tree(training, validation, test, output):
training_data = np.loadtxt(training, delimiter=',')
validation_data = np.loadtxt(validation, delimiter=',')
test_data = np.loadtxt(test, delimiter=',')
dt = DecisionTreeClassifier()
dt.fit(training_data[:, :-1], training_data[:, -1])
params = {'criterion': ['gini', 'entropy'],
'max_depth': [10, None],
'min_samples_split': [0.01, 0.1, 0.5, 0.9, 1.0],
'min_impurity_decrease': [0.0, 0.01, 0.1, 0.5, 0.9, 1.0],
'class_weight': [None, 'balanced']}
tuned_model = GridSearchCV(estimator=dt, param_grid=params, n_jobs=-1, cv=3)
tuned_model.fit(validation_data[:, :-1], validation_data[:, -1])
print(tuned_model.best_params_)
test_prediction = tuned_model.best_estimator_.predict(test_data[:, :-1])
helper.output_computed_metrics(test_data, test_prediction, output)
def best_mlp(training, validation, test, output):
training_data = np.loadtxt(training, delimiter=',')
validation_data = np.loadtxt(validation, delimiter=',')
test_data = np.loadtxt(test, delimiter=',')
mlp = MLPClassifier()
mlp.fit(training_data[:, :-1], training_data[:, -1])
params = {
'activation': ['logistic', 'tanh', 'relu', 'identity'],
'hidden_layer_sizes': [(30, 50), (10, 10, 10)],
'solver': ['adam', 'sgd']
}
tuned_model = GridSearchCV(estimator=mlp,
param_grid=params,
n_jobs=-1,
cv=3)
tuned_model.fit(validation_data[:, :-1], validation_data[:, -1])
print(tuned_model.best_params_)
test_prediction = tuned_model.best_estimator_.predict(test_data[:, :-1])
helper.output_computed_metrics(test_data, test_prediction, output)