def test_disttype(self, bmds3_complete_continuous):
data = deepcopy(bmds3_complete_continuous)
data["models"] = {
"frequentist_restricted": ["Hill", "Linear", "Power"],
}
# normal
data["options"][0]["dist_type"] = 1
session = AnalysisSession.create(data, 0, 0)
assert len(session.frequentist.models) == 3
names = [model.name() for model in session.frequentist.models]
assert names == ["Hill", "Linear", "Power"]
data["options"][0]["dist_type"] = 2
session = AnalysisSession.create(data, 0, 0)
assert len(session.frequentist.models) == 3
names = [model.name() for model in session.frequentist.models]
assert names == ["Hill", "Linear", "Power"]
# lognormal
data["options"][0]["dist_type"] = 3
session = AnalysisSession.create(data, 0, 0)
assert len(session.frequentist.models) == 1
names = [model.name() for model in session.frequentist.models]
assert names == ["Hill"]
def test_default_continuous_individual(
self, bmds3_complete_continuous_individual):
# assure a default dataset can be created
data = deepcopy(bmds3_complete_continuous_individual)
session = AnalysisSession.create(data, 0, 0)
assert len(session.frequentist.models) == 1
assert len(session.bayesian.models) == 1
def test_polynomial_unpacking(self, bmds3_complete_continuous):
# test linear; degree 0
data = deepcopy(bmds3_complete_continuous)
data["models"] = {"frequentist_unrestricted": ["Linear"]}
data["dataset_options"][0]["degree"] = 0
session = AnalysisSession.create(data, 0, 0)
assert len(session.frequentist.models) == 1
assert session.frequentist.models[0].settings.degree == 1
assert session.bayesian is None
# test polynomial; degree 2
data = deepcopy(bmds3_complete_continuous)
data["models"] = {"frequentist_unrestricted": ["Polynomial"]}
data["dataset_options"][0]["degree"] = 2
session = AnalysisSession.create(data, 0, 0)
assert len(session.frequentist.models) == 1
assert session.frequentist.models[0].settings.degree == 2
assert session.bayesian is None
# test polynomial; degree 3
data = deepcopy(bmds3_complete_continuous)
data["models"] = {"frequentist_unrestricted": ["Polynomial"]}
data["dataset_options"][0]["degree"] = 3
session = AnalysisSession.create(data, 0, 0)
assert len(session.frequentist.models) == 2
assert session.frequentist.models[0].settings.degree == 2
assert session.frequentist.models[1].settings.degree == 3
assert session.bayesian is None
# test linear + polynomial; degree 3
data = deepcopy(bmds3_complete_continuous)
data["models"] = {"frequentist_unrestricted": ["Linear", "Polynomial"]}
data["dataset_options"][0]["degree"] = 3
session = AnalysisSession.create(data, 0, 0)
assert len(session.frequentist.models) == 3
assert session.frequentist.models[0].settings.degree == 1
assert session.frequentist.models[1].settings.degree == 2
assert session.frequentist.models[2].settings.degree == 3
assert session.bayesian is None
def test_exponential_unpacking(self, bmds3_complete_continuous):
data = deepcopy(bmds3_complete_continuous)
data["models"] = {
"frequentist_restricted": ["Exponential"],
"bayesian": [{
"model": "Exponential",
"prior_weight": 1
}],
}
session = AnalysisSession.create(data, 0, 0)
assert len(session.frequentist.models) == 2
assert session.frequentist.models[
0].bmd_model_class.id == ContinuousModelIds.c_exp_m3
assert session.frequentist.models[
1].bmd_model_class.id == ContinuousModelIds.c_exp_m5
assert len(session.bayesian.models) == 2
assert session.bayesian.models[
0].bmd_model_class.id == ContinuousModelIds.c_exp_m3
assert session.bayesian.models[
1].bmd_model_class.id == ContinuousModelIds.c_exp_m5
def test_prior_classes(self, bmds3_complete_dichotomous):
# assure a default dataset can be created
data = deepcopy(bmds3_complete_dichotomous)
data["models"] = {
"frequentist_restricted": ["Gamma"],
"frequentist_unrestricted": ["Gamma"],
"bayesian": [{
"model": "Gamma",
"prior_weight": 1
}],
}
session = AnalysisSession.create(data, 0, 0)
assert len(session.frequentist.models) == 2
assert len(session.bayesian.models) == 1
assert (session.frequentist.models[0].settings.priors.prior_class is
PriorClass.frequentist_restricted)
assert (session.frequentist.models[1].settings.priors.prior_class is
PriorClass.frequentist_unrestricted)
assert session.bayesian.models[
0].settings.priors.prior_class is PriorClass.bayesian
def test_default_dichotomous(self, bmds3_complete_dichotomous):
# assure a default dataset can be created
data = deepcopy(bmds3_complete_dichotomous)
session = AnalysisSession.create(data, 0, 0)
assert len(session.frequentist.models) == 1
assert len(session.bayesian.models) == 1
def test_multistage_permutations(self, bmds3_complete_dichotomous):
def _expected_degree(session, n: int):
assert session.bayesian is None
assert len(session.frequentist.models) == n
model_classes = set([
model.bmd_model_class.id
for model in session.frequentist.models
])
assert model_classes == {DichotomousModelIds.d_multistage}
degrees = set([
model.settings.degree for model in session.frequentist.models
])
assert degrees == set(list(range(1, n + 1)))
# degree = 1
data = deepcopy(bmds3_complete_dichotomous)
data["models"] = {"frequentist_restricted": ["Multistage"]}
data["dataset_options"][0]["degree"] = 1
session = AnalysisSession.create(data, 0, 0)
_expected_degree(session, 1)
# degree = 2
data = deepcopy(bmds3_complete_dichotomous)
data["models"] = {"frequentist_restricted": ["Multistage"]}
data["dataset_options"][0]["degree"] = 2
session = AnalysisSession.create(data, 0, 0)
_expected_degree(session, 2)
# 3 dose-groups; degree = N-1; expected 2
for num_doses in range(3, 8):
expected_degree = min(max(num_doses - 1, 2), 8)
data = deepcopy(bmds3_complete_dichotomous)
data["datasets"] = [{
"dtype": "D",
"metadata": {
"id": 123
},
"doses": list(range(num_doses)),
"ns": [10] * num_doses,
"incidences": list(range(1, num_doses + 1)),
}]
assert len(data["datasets"][0]["doses"]) == num_doses
data["models"] = {"frequentist_restricted": ["Multistage"]}
data["dataset_options"][0]["degree"] = 0 # n-1
session = AnalysisSession.create(data, 0, 0)
print(f"{num_doses=} {expected_degree=}")
_expected_degree(session, expected_degree)
# degree = N -1, bayesian, fixed at degree == 2
data = deepcopy(bmds3_complete_dichotomous)
data["models"] = {
"bayesian": [{
"model": "Multistage",
"prior_weight": 1
}]
}
data["dataset_options"][0]["degree"] = 0
session = AnalysisSession.create(data, 0, 0)
assert session.frequentist is None
assert len(session.bayesian.models) == 1
model = session.bayesian.models[0]
assert model.bmd_model_class.id == DichotomousModelIds.d_multistage
assert model.settings.degree == 2