predictions = pd.read_csv(os.path.join(data_dir, 'test.csv.out'), header=None)
predictions = [round(num) for num in predictions.squeeze().values]
#evaluate model
accuracy_score(test_y, predictions)