def test_estimate_alpha_beta_biased():
n = 100
random_state = np.random.RandomState(seed=42)
X = random_state.rand(n, 3)
dis = euclidean_distances(X)
bias = 0.1 + random_state.rand(n)
bias = bias.reshape(n, 1)
beta_true, alpha_true = 1., -3.5
counts = beta_true * dis**alpha_true
counts *= bias * bias.T
counts[np.isnan(counts) | np.isinf(counts)] = 0
counts[np.arange(len(counts)), np.arange(len(counts))] = 0
counts = sparse.coo_matrix(np.triu(counts))
alpha, beta = negative_binomial.estimate_alpha_beta(
counts,
X,
bias=bias,
random_state=random_state,
use_zero_entries=True,
infer_beta=False)
assert_almost_equal(alpha_true, alpha, 3)
assert_almost_equal(beta_true, beta, 3)
alpha, beta = negative_binomial.estimate_alpha_beta(
sparse.coo_matrix(counts),
X,
bias=bias,
random_state=random_state,
infer_beta=False,
use_zero_entries=True)
assert_almost_equal(alpha_true, alpha, 3)
assert_almost_equal(beta_true, beta, 4)
def test_estimate_alpha_beta():
n = 100
random_state = np.random.RandomState(seed=42)
X = 5 * random_state.rand(n, 3)
dis = euclidean_distances(X)
beta_true, alpha_true = 2., -3.5
counts = beta_true * dis**alpha_true
counts[np.isnan(counts) | np.isinf(counts)] = 0
counts[np.arange(len(counts)), np.arange(len(counts))] = 0
counts = sparse.coo_matrix(np.triu(counts))
alpha, beta = negative_binomial.estimate_alpha_beta(
sparse.coo_matrix(counts), X, random_state=random_state)
assert_almost_equal(alpha_true, alpha, 3)
assert_almost_equal(beta_true, beta, 4)
def test_estimate_alpha_beta_biased():
n = 100
random_state = np.random.RandomState(seed=42)
X = random_state.rand(n, 3)
dis = euclidean_distances(X)
bias = 0.1 + random_state.rand(n)
bias = bias.reshape(n, 1)
beta_true, alpha_true = 2., -3.5
counts = beta_true * dis ** alpha_true
counts *= bias * bias.T
counts[np.isnan(counts) | np.isinf(counts)] = 0
counts[np.arange(len(counts)), np.arange(len(counts))] = 0
counts = sparse.coo_matrix(np.triu(counts))
alpha, beta = negative_binomial.estimate_alpha_beta(
counts, X, bias=bias, random_state=random_state)
assert_almost_equal(alpha_true, alpha, 3)
assert_almost_equal(beta_true, beta, 3)
alpha, beta = negative_binomial.estimate_alpha_beta(
sparse.coo_matrix(counts), X, bias=bias,
random_state=random_state)
assert_almost_equal(alpha_true, alpha, 5)
assert_almost_equal(beta_true, beta, 5)
def test_estimate_alpha_beta_dispersion():
n = 100
random_state = np.random.RandomState(seed=42)
X = 5 * random_state.rand(n, 3)
dis = euclidean_distances(X)
beta_true, alpha_true = 1., -3.5
counts = beta_true * dis**alpha_true
counts[np.isnan(counts) | np.isinf(counts)] = 0
counts[np.arange(len(counts)), np.arange(len(counts))] = 0
counts = sparse.coo_matrix(np.triu(counts))
alpha, beta = negative_binomial.estimate_alpha_beta(
counts,
X,
random_state=random_state,
infer_beta=False,
use_zero_entries=True)
assert_almost_equal(alpha_true, alpha, 3)