def test_bayes(self):
mu = Beta(n_ones=np.ones(1), n_zeros=np.ones(1))
model = Bernoulli(prob=mu)
self.assertEqual(repr(model),
"Bernoulli(prob=Beta(n_ones=[ 1.], n_zeros=[ 1.]))")
model.bayes(np.array([1., 1., 0.])[:, None])
self.assertEqual(repr(model),
"Bernoulli(prob=Beta(n_ones=[ 3.], n_zeros=[ 2.]))")
def test_pdf(self):
b = Bernoulli(np.ones(2))
self.assertTrue(
np.allclose(b.pdf(np.ones((3, 2))), 1)
)
self.assertTrue(
np.allclose(b.pdf(np.zeros((4, 2))), 0)
)
def test_init(self):
b = Bernoulli()
self.assertTrue(b.prob is None)
b = Bernoulli(prob=np.ones(3))
self.assertTrue(b.ndim == 1)
self.assertTrue(b.size == 3)
self.assertTrue(b.shape == (3, ))
self.assertTrue(np.allclose(b.prob, np.ones(3)))
b = Bernoulli(prob=0.5)
self.assertEqual(b.ndim, 0)
self.assertEqual(b.size, 1)
self.assertEqual(b.shape, ())
self.assertEqual(b.prob, 0.5)
def test_ml(self):
b = Bernoulli()
b.ml(np.array([0., 1., 1., 1.]))
self.assertEqual(b.prob, 0.75)
b = Bernoulli()
b.ml(np.ones((4, 5, 2)))
self.assertTrue(b.shape == (5, 2))
self.assertTrue(np.allclose(b.prob, np.ones((5, 2))))
def test_draw(self):
b = Bernoulli(np.ones((6, 2)))
self.assertTrue(
np.allclose(b.draw(), np.ones((1, 6, 2)))
)
b = Bernoulli(np.zeros((2, 3)))
self.assertTrue(
np.allclose(b.draw(4), np.zeros((4, 2, 3)))
)
def test_map(self):
mu = Beta(n_ones=np.ones(1), n_zeros=np.ones(1))
model = Bernoulli(prob=mu)
model.map(np.array([1., 1., 0.])[:, None])
self.assertTrue((model.prob == 2 / 3))
def test_var(self):
b = Bernoulli()
self.assertEqual(b.var, None)
b = Bernoulli(np.ones(3) * 0.5)
self.assertTrue(np.allclose(b.var, np.ones(3) * 0.25))
def test_mean(self):
b = Bernoulli()
self.assertEqual(b.mean, None)
b = Bernoulli(0.75)
self.assertEqual(b.mean, 0.75)
def test_repr(self):
b = Bernoulli()
self.assertEqual(repr(b), "Bernoulli(prob=None)")
b = Bernoulli(prob=np.zeros(5))
self.assertEqual(repr(b), "Bernoulli(prob=[ 0. 0. 0. 0. 0.])")