Test invariance

test_invariance.py

@pytest.fixture
def dummy_passengers():
    # Based on passenger 2 (high passenger class female)
    passenger2 = {'PassengerId': 2,
                  'Pclass': 1,
                  'Name': ' Mrs. John',
                  'Sex': 'female',
                  'Age': 38.0,
                  'SibSp': 1,
                  'Parch': 0,
                  'Ticket': 'PC 17599',
                  'Fare': 71.2833,
                  'Embarked': 'C'}

    return passenger2

def test_dt_invariance(dummy_titanic_dt, dummy_passengers):
    model = dummy_titanic_dt
    _, p2 = dummy_passengers

    # Get original survival probability of passenger 2
    test_df = pd.DataFrame.from_dict([p2], orient='columns')
    X, y = get_feats_and_labels(prep_df(test_df))
    p2_prob = model.predict(X)[0]  # 1.0

    # Change name from John to Berns (without changing gender or title)
    p2_name = p2.copy()
    p2_name['Name'] = ' Mrs. Berns'
    test_df = pd.DataFrame.from_dict([p2_name], orient='columns')
    X, y = get_feats_and_labels(prep_df(test_df))
    p2_name_prob = model.predict(X)[0]  # 1.0 

    # Change ticket number from 'PC 17599' to 'A/5 21171'
    p2_ticket = p2.copy()
    p2_ticket['ticket'] = 'A/5 21171'
    test_df = pd.DataFrame.from_dict([p2_ticket], orient='columns')
    X, y = get_feats_and_labels(prep_df(test_df))
    p2_ticket_prob = model.predict(X)[0]  # 1.0

    assert p2_prob == p2_name_prob == p2_ticket_prob