def cross_corpus(train_set_name, test_set_name):
# open data set and train model
train_set = DataSet(train_set_name)
x_train, y_train = train_set.load_training_data()
model = train_model(x_train, y_train)
# open test set and predict labels
test_set = DataSet(test_set_name)
x_test, y_test = test_set.load_training_data()
predict = model.predict(x_test)
print_classification_matrix(predict, y_test)
return model
def cross_validation(data_set_name):
# open data set
data_set = DataSet(data_set_name)
# get train and test set
x, y = data_set.load_training_data()
x_train, x_test, y_train, y_test = train_test_split(x,
y,
test_size=0.1,
shuffle=True)
# train and save model
model = train_model(x_train, y_train)
# get cross validation performance
predict = model.predict(x_test)
accuracy = accuracy_score(y_test, predict)
print('Cross validation accuracy:', accuracy)
# print_classification_matrix(predict, y_test)
return accuracy, y_test, predict
def leave_one_out_score(data_set_name):
# open data set
data_set = DataSet(data_set_name)
x, y = data_set.load_training_data()
correct_count = 0
for i in range(len(y)):
x_train = np.delete(x, i, axis=0)
y_train = np.delete(y, i)
x_test = [x[i]]
y_test = [y[i]]
# train model
model = train_model(x_train, y_train)
if model.predict(x_test)[0] == y_test:
correct_count += 1
print(i, 'correct,', correct_count / (i + 1))
else:
print(i, 'wrong,', correct_count / (i + 1))
print('accuracy:', correct_count / len(y))
# evaluate model on test data set (cross corpus)
x, y = test_data_set.load_training_data()
score = cross_val_score(model, x, y, scoring='neg_mean_squared_error')
print('Mean squared error: {}'.format(-score))
# plot
predict = model.predict(x)
plt.scatter(y, predict)
plt.show()
if __name__ == '__main__':
data_set = DataSet('cepp')
x, y = data_set.load_training_data()
x_train, x_test, y_train, y_test = train_test_split(x,
y,
test_size=0.4,
shuffle=True,
random_state=0)
regr = linear_model.LinearRegression(normalize=True)
#regr = linear_model.Ridge(alpha=0.001, normalize=True)
regr.fit(x_train, y_train)
predict = regr.predict(x_test)
scores = cross_val_score(regr,
x_test,