def test_bool():
assert get_bool({"v": True}, "v")
assert get_bool({"v": True}, "a") is False
with pytest.raises(TypeError, match=r".not a bool"):
get_bool({"v": "True"}, "v")
with pytest.raises(TypeError, match=r".not a bool"):
get_bool({"v": 1}, "v")
with pytest.raises(TypeError, match=r".not a bool"):
get_bool({"v": 1.0}, "v")
def setup_covariance_estimator(
cov_estimators: CovarianceManager,
cov_type: str,
y: NDArray,
x: NDArray,
params: NDArray,
entity_ids: NDArray,
time_ids: NDArray,
*,
debiased: bool = False,
extra_df: int = 0,
**cov_config: Any,
) -> Union[HomoskedasticCovariance]:
estimator = cov_estimators[cov_type]
kernel = get_string(cov_config, "kernel")
bandwidth = get_float(cov_config, "bandwidth")
group_debias = get_bool(cov_config, "group_debias")
clusters = get_array_like(cov_config, "clusters")
if estimator is HomoskedasticCovariance:
return HomoskedasticCovariance(y,
x,
params,
entity_ids,
time_ids,
debiased=debiased,
extra_df=extra_df)
elif estimator is HeteroskedasticCovariance:
return HeteroskedasticCovariance(y,
x,
params,
entity_ids,
time_ids,
debiased=debiased,
extra_df=extra_df)
elif estimator is ClusteredCovariance:
return ClusteredCovariance(
y,
x,
params,
entity_ids,
time_ids,
debiased=debiased,
extra_df=extra_df,
clusters=clusters,
group_debias=group_debias,
)
elif estimator is DriscollKraay:
return DriscollKraay(
y,
x,
params,
entity_ids,
time_ids,
debiased=debiased,
extra_df=extra_df,
kernel=kernel,
bandwidth=bandwidth,
)
else: # ACCovariance:
return ACCovariance(
y,
x,
params,
entity_ids,
time_ids,
debiased=debiased,
extra_df=extra_df,
kernel=kernel,
bandwidth=bandwidth,
)
def setup_covariance_estimator(
cov_estimators: CovarianceManager,
cov_type: str,
y: Float64Array,
x: Float64Array,
params: Float64Array,
entity_ids: IntArray,
time_ids: IntArray,
*,
debiased: bool = False,
extra_df: int = 0,
**cov_config: Any,
) -> Union[HomoskedasticCovariance]:
estimator = cov_estimators[cov_type]
unknown_kwargs = [
str(key) for key in cov_config
if str(key) not in estimator.ALLOWED_KWARGS
]
if unknown_kwargs:
if estimator.ALLOWED_KWARGS:
allowed = ", ".join(estimator.ALLOWED_KWARGS)
kwarg_err = f"only supports the keyword arguments: {allowed}"
else:
kwarg_err = "does not support any keyword arguments"
msg = (
f"Covariance estimator {estimator.__name__} {kwarg_err}. Unknown keyword "
f"arguments were passed to the estimator. The unknown keyword argument(s) "
f"are: {', '.join(unknown_kwargs)} ")
raise ValueError(msg)
kernel = get_string(cov_config, "kernel")
bandwidth = get_float(cov_config, "bandwidth")
group_debias = get_bool(cov_config, "group_debias")
clusters = get_array_like(cov_config, "clusters")
if estimator is HomoskedasticCovariance:
return HomoskedasticCovariance(y,
x,
params,
entity_ids,
time_ids,
debiased=debiased,
extra_df=extra_df)
elif estimator is HeteroskedasticCovariance:
return HeteroskedasticCovariance(y,
x,
params,
entity_ids,
time_ids,
debiased=debiased,
extra_df=extra_df)
elif estimator is ClusteredCovariance:
return ClusteredCovariance(
y,
x,
params,
entity_ids,
time_ids,
debiased=debiased,
extra_df=extra_df,
clusters=clusters,
group_debias=group_debias,
)
elif estimator is DriscollKraay:
return DriscollKraay(
y,
x,
params,
entity_ids,
time_ids,
debiased=debiased,
extra_df=extra_df,
kernel=kernel,
bandwidth=bandwidth,
)
else: # ACCovariance:
return ACCovariance(
y,
x,
params,
entity_ids,
time_ids,
debiased=debiased,
extra_df=extra_df,
kernel=kernel,
bandwidth=bandwidth,
)