def test_bootstrapmoment_skewness() -> None:
moment = BootstrapMoment(numsamples=10000, rng=1234)
values = np.arange(100, dtype=float)
moment.fill(values)
assert abs(moment.skewness().nominal - _skewness(values)) < 0.01
assert abs(np.average(moment.skewness().samples) -
_skewness(values)) < 10.0
def test_bootstrapmoment_standard_distributions(
with_weights: bool,
generator: Callable[[], np.ndarray],
mu: float,
sigma: float,
skewness: float,
) -> None:
moment = BootstrapMoment(numsamples=100, rng=1234)
values = generator()
if with_weights:
weight = np.random.default_rng(91011).uniform(size=values.shape)
moment.fill(values, weight=weight)
else:
moment.fill(values)
tolerance = 0.01
assert moment.mean().nominal == pytest.approx(mu,
rel=tolerance,
abs=tolerance)
assert np.average(moment.mean().samples) == pytest.approx(mu,
rel=tolerance,
abs=tolerance)
assert moment.std().nominal == pytest.approx(sigma,
rel=tolerance,
abs=tolerance)
assert np.average(moment.std().samples) == pytest.approx(sigma,
rel=tolerance,
abs=tolerance)
assert moment.skewness().nominal == pytest.approx(skewness,
rel=tolerance,
abs=tolerance)
assert np.average(moment.skewness().samples) == pytest.approx(
skewness, rel=tolerance, abs=tolerance)
def test_bootstrapmoment_array_shapes() -> None:
moment = BootstrapMoment(numsamples=3, rng=1234)
values = np.arange(100, dtype=float)
moment.fill(values)
assert (moment.mean().samples.shape == moment.std().samples.shape ==
moment.variance().samples.shape == moment.skewness().samples.shape
== (3, ))