def info_for_analysis_factor(
self,
analysis_factor
) -> str:
"""
A string describing a given AnalysisFactor in the context of this graph.
"""
model = analysis_factor.prior_model
formatter = TextFormatter()
for path, prior in model.path_instance_tuples_for_class(
Prior,
ignore_children=True
):
name = path[-1]
related_factor_names = self._related_factor_names(
prior,
excluded_factor=analysis_factor
)
if len(related_factor_names) > 0:
name = f"{name} ({related_factor_names})"
path = path[:-1] + (name,)
formatter.add(
path,
self.variable_formatter(
prior
)
)
return f"{analysis_factor.name}\n\n{formatter.text}"
def info(self) -> str:
"""
Use the priors that make up the model_mapper to generate information on each
parameter of the overall model.
This information is extracted from each priors *model_info* property.
"""
formatter = TextFormatter()
for t in self.path_instance_tuples_for_class((Prior, float, tuple)):
formatter.add(t)
return formatter.text
def info_for_prior_factor(
self,
prior_factor: PriorFactor
) -> str:
"""
A string describing a given PriorFactor in the context of this graph.
"""
related_factor_names = self._related_factor_names(
variable=prior_factor.variable,
excluded_factor=prior_factor
)
formatter = TextFormatter()
formatter.add(
(f"{prior_factor.name} ({related_factor_names})",),
self.variable_formatter(
prior_factor.variable
)
)
return formatter.text
def info_for_hierarchical_factor(
self,
hierarchical_factor
):
distribution_model_info = self.info_for_analysis_factor(
hierarchical_factor
)
formatter = TextFormatter()
for factor in hierarchical_factor.factors:
related_factor_names = self._related_factor_names(
variable=factor.variable,
excluded_factor=factor
)
formatter.add(
(related_factor_names,),
self.variable_formatter(
factor.variable
)
)
return f"{distribution_model_info}\n\nDrawn Variables\n\n{formatter.text}"
def make_results_text(self, model_approx) -> str:
"""
Create a string describing the posterior values after this factor
during or after an EPOptimisation.
Parameters
----------
model_approx: EPMeanField
Returns
-------
A string containing the name of this factor with the names and
values of each associated variable in the mean field.
"""
arguments = {
prior: model_approx.mean_field[prior] for prior in self.prior_model.priors
}
updated_model = self.prior_model.gaussian_prior_model_for_arguments(arguments)
formatter = TextFormatter()
for path, prior in updated_model.path_priors_tuples:
formatter.add(path, prior.mean)
return f"{self.name}\n\n{formatter.text}"