def test_binary_classification_predict_on_predictor_instance():
np.random.seed(0)
df_titanic_train, df_titanic_test = utils.get_titanic_binary_classification_dataset()
ml_predictor = utils.train_basic_binary_classifier(df_titanic_train)
predictions = ml_predictor.predict(df_titanic_test)
test_score = accuracy_score(predictions, df_titanic_test.survived)
# Make sure our score is good, but not unreasonably good
print(test_score)
assert .77 < test_score < .805
def test_binary_classification_predict_proba_on_predictor_instance():
np.random.seed(0)
df_titanic_train, df_titanic_test = utils.get_titanic_binary_classification_dataset()
ml_predictor = utils.train_basic_binary_classifier(df_titanic_train)
#
predictions = ml_predictor.predict_proba(df_titanic_test)
predictions = [pred[1] for pred in predictions]
test_score = utils.calculate_brier_score_loss(df_titanic_test.survived, predictions)
# Make sure our score is good, but not unreasonably good
print(test_score)
assert -0.16 < test_score < -0.135