def test_draw(self):
d = Dirichlet(np.array([1., 2., 1.]))
self.assertEqual(d.draw().shape, (1, 3))
sample = d.draw(1000)
self.assertTrue(
np.allclose(np.mean(sample, axis=0), [0.25, 0.5, 0.25], 1e-1, 1e-1)
)
def test_var(self):
d = Dirichlet(np.array([1, 2, 3]))
var = np.array(
[[5, -2, -3],
[-2, 8, -6],
[-3, -6, 9]]
) / 36 / 7
self.assertTrue(np.allclose(var, d.var))
def test_bayes(self):
mu = Dirichlet(concentration=np.ones(3))
model = Categorical(prob=mu)
self.assertEqual(
repr(model),
"Categorical(prob=\nDirichlet(concentration=[ 1. 1. 1.])\n)")
model.bayes(np.array([[1., 0., 0.], [1., 0., 0.], [0., 1., 0.]]))
self.assertEqual(
repr(model),
"Categorical(prob=\nDirichlet(concentration=[ 3. 2. 1.])\n)")
def test_init(self):
d = Dirichlet(np.ones(3))
self.assertTrue((d.concentration == 1).all())
self.assertEqual(d.size, 3)
self.assertEqual(d.ndim, 1)
self.assertEqual(d.shape, (3,))
def test_pdf(self):
d = Dirichlet(np.ones(4))
self.assertTrue((d.pdf(np.random.uniform(size=(5, 4))) == 6).all())
def test_mean(self):
d = Dirichlet(np.ones(4))
self.assertTrue((d.mean == 0.25).all())
def test_repr(self):
d = Dirichlet(np.ones(3))
self.assertEqual(repr(d), "Dirichlet(concentration=[ 1. 1. 1.])")
def test_init(self):
d = Dirichlet(np.ones(3))
self.assertTrue((d.concentration == 1).all())
self.assertEqual(d.n_classes, 3)