def test_preprocess_data_scaling_mean(self):
"""Test that the input features have been correctly scaled to zero mean."""
model = BPLModel(TEST_DATA, X=TEST_FEATS)
stan_data = model._pre_process_data()
self.assertTrue(
np.allclose(stan_data["X"].mean(axis=0), np.zeros(stan_data["X"].shape[1]))
)
def test_preprocess_data_scaling_std(self):
"""Test that the input features have been correctly scaled to 0.5 standard deviation."""
model = BPLModel(TEST_DATA, X=TEST_FEATS)
stan_data = model._pre_process_data()
self.assertTrue(
np.allclose(stan_data["X"].std(axis=0),
0.5 * np.ones(stan_data["X"].shape[1])))
def test_preprocess_data_nofeats(self):
"""Test that stan data dictionary has the correct keys when no features are passed."""
model = BPLModel(TEST_DATA, X=None)
stan_data = model._pre_process_data()
self.assertTrue(
set(stan_data.keys())
== {"nteam", "nmatch", "home_team", "away_team", "home_goals", "away_goals"}
)
def test_preprocess_data_date(self):
"""Test that the correct date constraint is applied in preprocessing."""
model = BPLModel(TEST_DATA, X=TEST_FEATS)
stan_data = model._pre_process_data(max_date="2018-03-01")
df = TEST_DATA.copy()
df.loc[:, "date"] = pd.to_datetime(df["date"])
df = df[df["date"] <= "2018-03-01"]
n = len(df)
self.assertEqual(stan_data["nmatch"], n)