def test_sample_ppc_var_names(model_fixture):
model, observed = model_fixture
trace = pm.inference.utils.trace_to_arviz({
"model/sd":
tf.ones((10, 1), dtype="float32"),
"model/y":
tf.convert_to_tensor(observed[:, None]),
})
with pytest.raises(ValueError):
forward_sampling.sample_posterior_predictive(model(),
trace,
var_names=[])
with pytest.raises(KeyError):
forward_sampling.sample_posterior_predictive(
model(), trace, var_names=["name not in model!"])
with pytest.raises(TypeError):
trace.posterior["name not in model!"] = tf.constant(1.0)
pm.sample_posterior_predictive(model(), trace)
del trace.posterior["name not in model!"]
var_names = ["model/sd", "model/x", "model/dy"]
ppc = pm.sample_posterior_predictive(
model(), trace, var_names=var_names).posterior_predictive
assert set(var_names) == set(ppc)
assert ppc["model/sd"].shape == trace.posterior["model/sd"].shape
def test_sample_ppc_corrupt_trace():
@pm.model
def model():
x = yield pm.Normal("x", tf.ones(5), 1)
y = yield pm.Normal("y", x, 1)
trace1 = {"model/x": np.ones(7, dtype="float32")}
trace2 = {
"model/x": np.ones(5, dtype="float32"),
"model/y": np.array(0, dtype="float32")
}
with pytest.raises(EvaluationError):
pm.sample_posterior_predictive(model(), trace1)
with pytest.raises(EvaluationError):
pm.sample_posterior_predictive(model(), trace2)
def test_posterior_predictive(mixture):
model, n, _ = mixture
trace = pm.sample(model, num_samples=100, num_chains=2)
ppc = pm.sample_posterior_predictive(model, trace).posterior_predictive
if n == 1:
assert ppc["mixture/mixture"].shape == (2, 100, 100)
else:
assert ppc["mixture/mixture"].shape == (2, 100, 100, n)
def test_sample_ppc_var_names(model_fixture):
model, observed = model_fixture
trace = {
"model/sd": tf.convert_to_tensor(np.array(1.0, dtype="float32")),
"model/y": tf.convert_to_tensor(observed),
}
with pytest.raises(ValueError):
pm.sample_posterior_predictive(model(), trace, var_names=[])
with pytest.raises(KeyError):
pm.sample_posterior_predictive(model(),
trace,
var_names=["name not in model!"])
with pytest.raises(TypeError):
bad_trace = trace.copy()
bad_trace["name not in model!"] = tf.constant(1.0)
pm.sample_posterior_predictive(model(), bad_trace)
var_names = ["model/sd", "model/x", "model/dy"]
ppc = pm.sample_posterior_predictive(model(), trace, var_names=var_names)
assert set(var_names) == set(ppc)
assert ppc["model/sd"].shape == trace["model/sd"].shape
assert np.all(
[v.shape == observed.shape for k, v in ppc.items() if k != "model/sd"])
def test_sample_ppc_corrupt_trace():
@pm.model
def model():
x = yield pm.Normal("x", tf.ones(5), 1)
y = yield pm.Normal("y", x, 1)
trace1 = pm.inference.utils.trace_to_arviz(
{"model/x": tf.ones((7, 1), dtype="float32")})
trace2 = pm.inference.utils.trace_to_arviz({
"model/x":
tf.ones((1, 5), dtype="float32"),
"model/y":
tf.zeros((1, 1), dtype="float32")
})
with pytest.raises(EvaluationError):
pm.sample_posterior_predictive(model(), trace1)
with pytest.raises(EvaluationError):
pm.sample_posterior_predictive(model(), trace2)
def test_sample_posterior_predictive(posterior_predictive_fixture):
(
model,
observed,
core_ppc_shapes,
observed_in_RV,
trace,
num_samples,
num_chains,
) = posterior_predictive_fixture
if observed_in_RV:
observed_kwarg = None
else:
observed_kwarg = observed
ppc = pm.sample_posterior_predictive(
model(), trace, observed=observed_kwarg).posterior_predictive
assert set(sorted(list(ppc))) == set(observed)
assert np.all([
v.shape == (num_chains, num_samples) + observed[k].shape
for k, v in ppc.items()
])
def test_data_from_model(model, modelParams, params_dict, to_return=None):
"""
Generates a test dataset from our model! Needs some pregenerated
data file which is quite strange but only uses it to cast the right dimensions!
Parameters
----------
model: pymc4 model
The model to sample from
modelParams: :py:class:`covid19_npis.ModelParams`
Instance of modelParams, mainly used for number of age groups and
number of countries.
params_dict: dictionary
Parameters for the test run.
to_return: list str
Return these variables
Returns
-------
: new_cases_inferred, R_t, interventions
"""
# ------------------------------------------------------------------------------ #
# Set params for the test dataset
# ------------------------------------------------------------------------------ #
len_gen_interv_kernel = 12
num_interventions = 2
if to_return is None:
to_return = []
model_name = model(modelParams).name
dict_with_model_name = {
f"{model_name}/{key}": tf.cast(value, "float32")[tf.newaxis,
tf.newaxis]
for key, value in params_dict.items()
}
trace = az.from_dict(posterior=dict_with_model_name, )
var_names = []
# variables = list(
# set(
# # ["R_0", "new_cases_inferred", "R_t", "g", "d_i_c_p", "new_I_t", "h_0_t",]
# to_return
# )
# )
variables = to_return
for var in variables:
var_names.append(f"{model_name}/{var}")
# Sample
trace = pm.sample_posterior_predictive(
model(modelParams),
trace,
var_names=var_names,
use_auto_batching=False,
)
# Convert to pandas
_, sample_state = pm.evaluate_model(model(modelParams))
def convert_to_pandas(key):
df = data.convert_trace_to_dataframe(
trace,
sample_state=sample_state,
key=key,
data_type="posterior_predictive",
)
df.index = df.index.droplevel(["chain", "draw"])
if "time" in df.index.names:
df = df.stack().unstack(level="time").T
df.columns = df.columns.droplevel(
-1) # stack adds an additional dimension
df.index = df.index.rename("date")
return df
"""
new_cases_inferred = convert_to_pandas("new_cases_inferred")
R_t = convert_to_pandas("R_t")
d = data.convert_trace_to_dataframe(
trace,
sample_state=sample_state,
key="d_i_c_p",
data_type="posterior_predictive",
)
# Take intervention array from model params
for c, country in enumerate(modelParams.countries):
if c == 0:
interv = country.data_interventions
else:
interv = interv.join(country.data_interventions)
extra = []
for var in to_return:
extra.append(convert_to_pandas(var))
"""
dfs = [convert_to_pandas(var) for var in variables]
return dfs