def test_prior_predictive(crossed_data):
crossed_data["count"] = (crossed_data["Y"] -
crossed_data["Y"].min()).round()
# New default priors are too wide for this case... something to keep investigating
model = Model(
"count ~ threecats + continuous + dummy",
crossed_data,
family="poisson",
automatic_priors="mle",
)
model.fit(tune=0, draws=2)
pps = model.prior_predictive(draws=500)
keys = ["Intercept", "threecats", "continuous", "dummy"]
shapes = [(1, 500), (1, 500, 2), (1, 500), (1, 500)]
for key, shape in zip(keys, shapes):
assert pps.prior[key].shape == shape
assert pps.prior_predictive["count"].shape == (500, 120)
assert pps.observed_data["count"].shape == (120, )
pps = model.prior_predictive(draws=500, var_names=["count"])
assert pps.groups() == ["prior_predictive", "observed_data"]
pps = model.prior_predictive(draws=500, var_names=["Intercept"])
assert pps.groups() == ["prior"]
def test_prior_predictive(crossed_data):
crossed_data["count"] = (crossed_data["Y"] - crossed_data["Y"].min()).round()
model = Model("count ~ threecats + continuous + dummy", crossed_data, family="poisson")
model.fit(tune=0, draws=2)
pps = model.prior_predictive(draws=500)
keys = ["Intercept", "threecats", "continuous", "dummy"]
shapes = [(1, 500), (1, 500, 2), (1, 500), (1, 500)]
for key, shape in zip(keys, shapes):
assert pps.prior[key].shape == shape
assert pps.prior_predictive["count"].shape == (1, 500, 120)
assert pps.observed_data["count"].shape == (120,)
pps = model.prior_predictive(draws=500, var_names=["count"])
assert pps.groups() == ["prior_predictive", "observed_data"]
pps = model.prior_predictive(draws=500, var_names=["Intercept"])
assert pps.groups() == ["prior"]
