def test_kde(self):
zikm = eda.ZeroIGKdeMdl(self.x)
kde = zikm.kde
kde.logpdf([1, 2, 3])
kde2 = eda.ZeroIGKdeMdl(self.x_all_zero).kde
assert kde2 is None
kde3 = eda.ZeroIGKdeMdl(self.x_one_nonzero).kde
assert kde3 is None
def test_kde_bw(self):
zikm = eda.ZeroIGKdeMdl(self.x)
n = len(zikm.x_nonzero)
d = 1
# scott: n**(-1./(d+4))
np.testing.assert_allclose(
zikm.bandwidth,
zikm.x_nonzero.std(ddof=1) * n**(-1 / (d + 4)))
# silverman: (n * (d + 2) / 4.)**(-1. / (d + 4))
zikm2 = eda.ZeroIGKdeMdl(self.x, "silverman")
np.testing.assert_allclose(
zikm2.bandwidth,
zikm2.x_nonzero.std(ddof=1) * (n * (d + 2) / 4.)**(-1. / (d + 4)))
zikm3 = eda.ZeroIGKdeMdl(self.x, 1)
xnz_std = zikm.x_nonzero.std(ddof=1)
np.testing.assert_allclose(1, zikm3.bandwidth / xnz_std)
assert not np.allclose(zikm2.bandwidth, zikm3.bandwidth)
def test_all_zero(self):
zikm = eda.ZeroIGKdeMdl(self.x_all_zero)
assert zikm.bandwidth is None
np.testing.assert_allclose(zikm.zi_mdl,
np.log(3) + np.log(len(self.x_all_zero)))
assert zikm.x_nonzero.size == 0
np.testing.assert_allclose(zikm.x, self.x_all_zero)
np.testing.assert_allclose(zikm.kde_mdl, 0)
np.testing.assert_allclose(zikm.mdl, zikm.zi_mdl + zikm.kde_mdl)
def test_encode(self):
zimdl = eda.ZeroIGKdeMdl(self.x)
qx = [0, 0, 0] + [1, 2, 10, 20]
emdl = zimdl.encode(qx)
ezi = zimdl._zi_encoder.encode(qx)
ekde = zimdl._kde_encoder.encode(list(filter(lambda x: x != 0, qx)))
qp = zimdl._k / zimdl._n
np.testing.assert_allclose(
ezi, -np.log(qp) * 4 - np.log(1 - qp) * 3 + np.log(3))
np.testing.assert_allclose(emdl, ezi + ekde)
def test_std_usage(self):
zikm = eda.ZeroIGKdeMdl(self.x)
np.testing.assert_allclose(zikm.x, self.x)
assert zikm.x is not self.x
np.testing.assert_allclose(zikm.x_nonzero, self.x[50:250])
np.testing.assert_allclose(zikm.mdl, zikm.zi_mdl + zikm.kde_mdl)
np.testing.assert_allclose(
zikm.zi_mdl,
np.log(3) + eda.MultinomialMdl(self.x != 0).mdl)
assert zikm._bw_method == "scott"
assert zikm.bandwidth is not None
# test > 0 value kde same
zikm2 = eda.ZeroIGKdeMdl(self.x[50:250])
np.testing.assert_allclose(zikm2.kde_mdl, zikm.kde_mdl)
np.testing.assert_allclose(zikm2.bandwidth, zikm.bandwidth)
def test_wrong_x_shape(self):
with pytest.raises(ValueError) as excinfo:
eda.ZeroIGKdeMdl(np.arange(10).reshape(5, 2))
def test_empty(self):
zikm = eda.ZeroIGKdeMdl(np.array([]))
assert zikm.mdl == np.log(3)
assert zikm.zi_mdl == np.log(3)
assert zikm.kde_mdl == 0
def test_one_nonzero(self):
zikm = eda.ZeroIGKdeMdl(self.x_one_nonzero)
assert zikm.bandwidth is None
# 1 non-0, kde falls back on uniform
np.testing.assert_allclose(
zikm.kde_mdl, np.log(np.max(np.abs(self.x_one_nonzero)) * 2))
def test_all_nonzero(self):
zikm = eda.ZeroIGKdeMdl(self.x_all_non_zero)
np.testing.assert_allclose(
zikm.zi_mdl,
np.log(3) + np.log(len(self.x_all_non_zero)))