def test_creating_discrete_distribution_with_non_normalized_weights():
values = [10, 20, 30]
weights = [5, 0, 15]
disc = Discrete(values, weights)
np.testing.assert_almost_equal(disc.getp(10), 0.25)
np.testing.assert_almost_equal(disc.getp(20), 0)
np.testing.assert_almost_equal(disc.getp(30), 0.75)
def test_generate_produces_np_array_of_given_size(values, weights, size):
disc = Discrete(values, weights)
samples = disc.generate(size)
est_prob = [sum(samples == i) / len(samples) for i in values]
np.testing.assert_almost_equal(est_prob, disc.prob, 1)
assert set(samples) == set(values)
def test_calling_generates_with_given_pmf(values, weights, n):
disc = Discrete(values, weights)
samples = np.asarray([disc() for _ in range(n)])
est_prob = [sum(samples == i) / len(samples) for i in values]
np.testing.assert_almost_equal(est_prob, disc.prob, 1)
assert set(samples) == set(values)
def test_moment_estimation(values, weights, m1, m2, m3):
disc = Discrete(values, weights)
assert disc.moment(1) == m1
assert disc.moment(2) == m2
assert disc.moment(3) == m3
def test_creation_with_empty_values_raises_error():
with pytest.raises(ValueError) as excinfo:
Discrete([])
assert 'expected non-empty values' in str(excinfo.value).lower()
def test_mean_estimation(values, weights, expected):
disc = Discrete(values, weights)
assert disc.mean() == expected
def test_getp_returns_zero_for_any_nonexisting_value():
disc = Discrete([10, 20])
assert disc.getp(13) == 0
assert disc.getp(0) == 0
def test_creating_discrete_distribution_with_dictionary(values, probs):
disc = Discrete(values)
assert set(disc.values) == set(values.keys())
np.testing.assert_almost_equal(disc.prob, probs)
def test_values_get_equal_probability_if_weights_not_given(values):
disc = Discrete(values)
expected_p = np.asarray([1. / len(values)] * len(values))
np.testing.assert_almost_equal(disc.values, values)
np.testing.assert_almost_equal(disc.prob, expected_p)
def test_weights_sum_must_be_positive():
with pytest.raises(ValueError) as excinfo:
Discrete([10, 20], [0, 0])
assert 'weights sum must be positive' in str(excinfo.value).lower()
def test_weights_must_be_non_negative():
with pytest.raises(ValueError) as excinfo:
Discrete([10, 20], [0.5, -0.1])
assert 'weights must be non-negative' in str(excinfo.value).lower()
def test_values_and_weights_sizes_must_match(values, weights):
with pytest.raises(ValueError) as excinfo:
Discrete(values, weights=weights)
assert 'values and weights size mismatch' in str(excinfo.value).lower()
def test_calling_discrete_distribution_with_single_values_returns_it(value, n):
disc = Discrete([value])
samples = [disc() for _ in range(n)]
assert samples == [value] * n
def test_stddev_and_var(values, weights, var):
disc = Discrete(values, weights)
np.testing.assert_almost_equal(disc.var(), var)
np.testing.assert_almost_equal(disc.std(), var ** 0.5)
def test_negative_or_float_moment_parameter_raises_error(k):
disc = Discrete([10, 20, 30])
with pytest.raises(ValueError) as excinfo:
disc.moment(k)
assert 'positive integer expected' in str(excinfo.value).lower()