def bootstrap_one_branch(sc,
data,
stat_fn=np.mean,
num_samples=10000,
seed_start=None,
threshold_quantile=None,
summary_quantiles=mabs.DEFAULT_QUANTILES):
"""Run a bootstrap for one branch on its own.
Resamples the data ``num_samples`` times, computes ``stat_fn`` for
each sample, then returns summary statistics for the distribution
of the outputs of ``stat_fn``.
Args:
sc: The spark context
data: The data as a list, 1D numpy array, or pandas Series
stat_fn: Either a function that aggregates each resampled
population to a scalar (e.g. the default value ``np.mean``
lets you bootstrap means), or a function that aggregates
each resampled population to a dict of scalars. In both
cases, this function must accept a one-dimensional ndarray
as its input.
num_samples: The number of bootstrap iterations to perform
seed_start: An int with which to seed numpy's RNG. It must
be unique within this set of calculations.
threshold_quantile (float, optional): An optional threshold
quantile, above which to discard outliers. E.g. ``0.9999``.
summary_quantiles (list, optional): Quantiles to determine the
confidence bands on the branch statistics. Change these
when making Bonferroni corrections.
"""
samples = get_bootstrap_samples(sc, data, stat_fn, num_samples, seed_start,
threshold_quantile)
return mabs.summarize_one_branch_samples(samples, summary_quantiles)
def test_summarize_one_branch_samples():
s = pd.Series(np.linspace(0, 1, 1001))
res = mabs.summarize_one_branch_samples(s, [0.05, 0.31, 0.95])
assert res.shape == (4, )
assert res["0.05"] == pytest.approx(0.05)
assert res["0.31"] == pytest.approx(0.31)
assert res["0.95"] == pytest.approx(0.95)
assert res["mean"] == pytest.approx(0.5)
def test_summarize_one_branch_vs_samples():
s = pd.Series([30, 80], index=["num_conversions", "num_enrollments"])
ppf = mabsb.summarize_one_branch_from_agg(s, quantiles=[0.5, 0.41])
samples = mabsb.get_samples(s.to_frame().T, "num_enrollments",
"num_conversions", 100000)
res = mabs.summarize_one_branch_samples(samples.iloc[:, 0],
quantiles=[0.5, 0.41])
assert ppf["0.5"] == pytest.approx(res["0.5"], abs=0.001)
assert ppf["0.41"] == pytest.approx(res["0.41"], abs=0.001)
assert ppf["mean"] == pytest.approx(res["mean"], abs=0.001)
def test_summarize_one_branch_samples_batch():
s = pd.Series(np.linspace(0, 1, 1001))
df = pd.DataFrame({"a": s, "b": s + 1})
res = mabs.summarize_one_branch_samples(df, quantiles=[0.05, 0.31, 0.95])
assert res.shape == (2, 4)
assert res.loc["a", "0.05"] == pytest.approx(0.05)
assert res.loc["a", "0.31"] == pytest.approx(0.31)
assert res.loc["a", "0.95"] == pytest.approx(0.95)
assert res.loc["a", "mean"] == pytest.approx(0.5)
assert res.loc["b", "0.05"] == pytest.approx(1.05)
assert res.loc["b", "0.31"] == pytest.approx(1.31)
assert res.loc["b", "0.95"] == pytest.approx(1.95)
assert res.loc["b", "mean"] == pytest.approx(1.5)
def test_summarize_one_branch_samples_batch():
s = pd.Series(np.linspace(0, 1, 1001))
df = pd.DataFrame({'a': s, 'b': s + 1})
res = mabs.summarize_one_branch_samples(df, quantiles=[0.05, 0.31, 0.95])
assert res.shape == (2, 4)
assert res.loc['a', '0.05'] == pytest.approx(0.05)
assert res.loc['a', '0.31'] == pytest.approx(0.31)
assert res.loc['a', '0.95'] == pytest.approx(0.95)
assert res.loc['a', 'mean'] == pytest.approx(0.5)
assert res.loc['b', '0.05'] == pytest.approx(1.05)
assert res.loc['b', '0.31'] == pytest.approx(1.31)
assert res.loc['b', '0.95'] == pytest.approx(1.95)
assert res.loc['b', 'mean'] == pytest.approx(1.5)
def bootstrap_one_branch(
data,
stat_fn=bb_mean,
num_samples=10000,
seed_start=None,
threshold_quantile=None,
summary_quantiles=mabs.DEFAULT_QUANTILES,
sc=None,
):
"""Bootstrap ``stat_fn`` for one branch on its own.
Computes ``stat_fn`` for ``num_samples`` resamples of ``data``,
then returns summary statistics for the results.
Args:
data: The data as a list, 1D numpy array, or pandas Series
stat_fn (callable, optional): A function that either:
* Aggregates each resampled population to a scalar (e.g.
the default, ``bb_mean``), or
* Aggregates each resampled population to a dict of
scalars (e.g. the func returned by
``make_bb_quantile_closure`` when given multiple
quantiles.
In both cases, this function must accept two parameters:
* a one-dimensional ndarray or pandas Series of values,
* an identically shaped object of weights for these values
num_samples: The number of bootstrap iterations to perform
seed_start: An int with which to seed numpy's RNG. It must
be unique within this set of calculations.
threshold_quantile (float, optional): An optional threshold
quantile, above which to discard outliers. E.g. ``0.9999``.
summary_quantiles (list, optional): Quantiles to determine the
confidence bands on the branch statistics. Change these
when making Bonferroni corrections.
sc (optional): The Spark context, if available
"""
samples = get_bootstrap_samples(data, stat_fn, num_samples, seed_start,
threshold_quantile, sc)
return mabs.summarize_one_branch_samples(samples, summary_quantiles)