def test_wrong_update(self):
"""
Tests ValueError if the passing update rule is not correct.
"""
# Put in incorrect update rule.
expectation_maximization = EG(update_rule='SS')
with self.assertRaises(ValueError):
# Running allocate will raise ValueError.
expectation_maximization.allocate(self.data, resample_by='M')
def test_em_solution(self):
"""
Test calculation of exponential gradient weights with expectation maximization update rule.
"""
# Use expectation maximization update rule.
expectation_maximization = EG(update_rule='EM')
# Allocates asset prices to EM.
expectation_maximization.allocate(self.data, resample_by='M')
all_weights = np.array(expectation_maximization.all_weights)
for i in range(all_weights.shape[0]):
weights = all_weights[i]
assert (weights >= 0).all()
assert len(weights) == self.data.shape[1]
np.testing.assert_almost_equal(np.sum(weights), 1)
def test_mu_solution(self):
"""
Test calculation of exponential gradient weights with multiplicative update rule.
"""
# Use multiplicative update rule.
multiplicative_update = EG(update_rule='MU')
# Allocates asset prices to MU.
multiplicative_update.allocate(self.data, resample_by='M')
all_weights = np.array(multiplicative_update.all_weights)
for i in range(all_weights.shape[0]):
weights = all_weights[i]
assert (weights >= 0).all()
assert len(weights) == self.data.shape[1]
np.testing.assert_almost_equal(np.sum(weights), 1)