def test_predict_proba_raises_error_if_not_fit(self):
with self.assertRaises(PyMC3ModelsError) as no_fit_error:
test_HLR = HierarchicalLogisticRegression()
test_HLR.predict_proba(self.X_train, self.cat_train)
expected = 'Run fit on the model before predict.'
self.assertEqual(str(no_fit_error.exception), expected)
class HierarchicalLogisticRegressionTestCase(unittest.TestCase):
def setUp(self):
def numpy_invlogit(x):
return 1 / (1 + np.exp(-x))
self.num_cats = 3
self.num_pred = 1
self.num_samples_per_cat = 100000
# Set random seed for repeatability
np.random.seed(27)
self.alphas = np.random.randn(self.num_cats)
self.betas = np.random.randn(self.num_cats, self.num_pred)
# TODO: make this more efficient; right now, it's very explicit.
x_a = np.random.randn(self.num_samples_per_cat, self.num_pred)
y_a = np.random.binomial(
1, numpy_invlogit(self.alphas[0] + np.sum(self.betas[0] * x_a, 1)))
x_b = np.random.randn(self.num_samples_per_cat, self.num_pred)
y_b = np.random.binomial(
1, numpy_invlogit(self.alphas[1] + np.sum(self.betas[1] * x_b, 1)))
x_c = np.random.randn(self.num_samples_per_cat, self.num_pred)
y_c = np.random.binomial(
1, numpy_invlogit(self.alphas[2] + np.sum(self.betas[2] * x_c, 1)))
X = np.concatenate([x_a, x_b, x_c])
Y = np.concatenate([y_a, y_b, y_c])
cats = np.concatenate([
np.zeros(self.num_samples_per_cat, dtype=np.int),
np.ones(self.num_samples_per_cat, dtype=np.int),
2 * np.ones(self.num_samples_per_cat, dtype=np.int)
])
output = train_test_split(X, cats, Y, test_size=0.4)
self.X_train, self.X_test, self.cat_train, self.cat_test, self.Y_train, self.Y_test = output
self.test_HLR = HierarchicalLogisticRegression()
# Fit the model once
inference_args = {
'n': 60000,
'callbacks': [pm.callbacks.CheckParametersConvergence()]
}
# Note: print is here so PyMC3 output won't overwrite the test name
print('')
self.test_HLR.fit(self.X_train,
self.Y_train,
self.cat_train,
num_advi_sample_draws=5000,
minibatch_size=2000,
inference_args=inference_args)
self.test_dir = tempfile.mkdtemp()
def tearDown(self):
shutil.rmtree(self.test_dir)
def setUp(self):
def numpy_invlogit(x):
return 1 / (1 + np.exp(-x))
self.num_cats = 3
self.num_pred = 1
self.num_samples_per_cat = 100000
self.alphas = np.random.randn(self.num_cats)
self.betas = np.random.randn(self.num_cats, self.num_pred)
#TODO: make this more efficient; right now, it's very explicit so I understand it.
x_a = np.random.randn(self.num_samples_per_cat, self.num_pred)
y_a = np.random.binomial(
1, numpy_invlogit(self.alphas[0] + np.sum(self.betas[0] * x_a, 1)))
x_b = np.random.randn(self.num_samples_per_cat, self.num_pred)
y_b = np.random.binomial(
1, numpy_invlogit(self.alphas[1] + np.sum(self.betas[1] * x_b, 1)))
x_c = np.random.randn(self.num_samples_per_cat, self.num_pred)
y_c = np.random.binomial(
1, numpy_invlogit(self.alphas[2] + np.sum(self.betas[2] * x_c, 1)))
X = np.concatenate([x_a, x_b, x_c])
Y = np.concatenate([y_a, y_b, y_c])
cats = np.concatenate([
np.zeros(self.num_samples_per_cat, dtype=np.int),
np.ones(self.num_samples_per_cat, dtype=np.int),
2 * np.ones(self.num_samples_per_cat, dtype=np.int)
])
self.X_train, self.X_test, self.cat_train, self.cat_test, self.Y_train, self.Y_test = train_test_split(
X, cats, Y, test_size=0.4)
self.test_HLR = HierarchicalLogisticRegression()
self.test_dir = tempfile.mkdtemp()
def test_save_and_load_work_correctly(self):
probs1 = self.test_HLR.predict_proba(self.X_test, self.cat_test)
self.test_HLR.save(self.test_dir)
HLR2 = HierarchicalLogisticRegression()
HLR2.load(self.test_dir)
self.assertEqual(self.test_HLR.num_cats, HLR2.num_cats)
self.assertEqual(self.test_HLR.num_pred, HLR2.num_pred)
self.assertEqual(self.test_HLR.num_training_samples,
HLR2.num_training_samples)
pd.testing.assert_frame_equal(summary(self.test_HLR.trace),
summary(HLR2.trace))
probs2 = HLR2.predict_proba(self.X_test, self.cat_test)
np.testing.assert_almost_equal(probs2, probs1, decimal=1)