credibility.report¶

The outputs of a credibility assessment: the record, the tables and the report.

write_outputs writes an assessment into a directory of its own:

• credibility.json: the assessment, see CredibilityAssessment.to_json.
• contexts.tsv: one row per context of use with its level, influence, consequence, risk, verdict, the decisive and the evaluable points, the minimum of evaluable points, the uncovered quantities, whether it is verified and per acceptance rule, of the context and of the evidence extensions, its value (<name>) and limit (<name>_limit).
• evidence.tsv: one row per context of use and evidence and per limitation (the limitations of the model with the context ""): the role, the points (n), mappings and studies, GMFE, AFE and the criteria.
• domain.tsv: one row per context of use, field and value of the domain: whether it is claimed, its coverage and its points (n).
• quantities.tsv: one row per context of use and quantity of interest: its description, whether it is a ratio, its coverage and its points.
• verification.tsv: one row per context of use and check of the verification, empty if the assessment was not verified: the factor, the compared points (n), the largest relative difference and its point, whether the acceptance changed, whether the check passed and its problems; the checks of the battery are rows of every context of use.
• uncertainty.tsv: one row per context of use and acceptance rule over the parameter draws of the adopted fits, empty without the propagation: the probability that the rule is met with its anytime-valid sequence, the decision of the statement at the threshold, the draws and the failed and needed draws.
• uncertainty_quantities.tsv: one row per context of use and quantity of interest which is no ratio: the points, the median fold width of the prediction bands and their coverage of the observations.
• uncertainty_sensitivity.tsv: one row per context of use, target (a quantity or a rule) and fitted parameter: the first-order index of the target over the parameter draws.
• variability.tsv: one row per context of use with the predictive check of its decisive points with the declared variability of the subjects, empty without: the subjects, the parameter vectors, the points and the means of groups, the coverage with its Wilson interval, the expected coverage and the agreement.
• variability_quantities.tsv: one row per context of use and quantity of interest which is no ratio: the points with a band and their coverage.
• vpc.tsv: one row per data point of the adopted fits with the band of its predictive check, empty without: the subjects of the observation, the statistic, the typical prediction, the quantiles and whether it is inside.
• cross_validation.tsv: one row per cross-validated adopted fit and held-out study: its points, the GMFE and the share within 2-fold of the held-out and of the fitted predictions, the cost of the refit.
• battery.tsv: one row per check of the battery of the variants, empty if it did not run: the factor, the status, the cases, the largest deviation and where, the statement and the problems.
• the tables of the evidence extensions (EvidenceExtension.report_tables) and of the battery hook (BatteryHook.report_tables), <name>.tsv.
• index.html: the report (render_html), which extends report_base.html of sbmlsim.report.templates and loads nothing from the network.

print_summary prints the table of the contexts of use, their verdicts, the role of the evidence which decides them and the evaluable points with their minimum, and how many of them are verified.

contexts_table(assessment)

evidence_table(assessment)

domain_table(assessment)

quantities_table(assessment)

verification_table(assessment)

uncertainty_table(assessment)

uncertainty_quantities_table(assessment)

sensitivity_table(assessment)

variability_table(assessment)

variability_quantities_table(assessment)

vpc_table(assessment)

cross_validation_table(assessment)

battery_table(assessment)

render_html(assessment, extensions=(), hook=None)

write_outputs(
    assessment, output_dir, extensions=(), hook=None
)

print_summary(assessment)