Creates a validated calibration_result S3 object. This is the
canonical output of both fit_calibration_freq() (curveRfreq) and
fit_calibration_bayes() (curveRbayes).
Usage
new_calibration_result(
meta,
ensemble = list(),
selection = list(),
grid = data.frame(),
samples = NULL,
diagnostics = NULL,
standards = NULL,
blanks = NULL,
population = NULL
)Arguments
- meta
Named list of metadata. Required fields:
method("frequentist"or"bayesian"),package,version,feature,antigen,plate,response_var,independent_var,is_log_response,is_log_independent.- ensemble
Named list of model fit results, one per model attempted. Each element should be a list with
model_name,converged,parameters,fit_stats, and optionallyraw_fit.- selection
Named list describing best-model selection:
best_model_name,criterion,weights(data.frame).- grid
Data frame of grid predictions from the best model.
- samples
Data frame of test sample predictions from the best model, or NULL if no samples were provided.
- diagnostics
Named list of diagnostic quantities (inflection point, LODs, LOQs, etc.), or NULL.
- standards
Data frame of standard data used to fit the curve, or NULL if not provided.
- blanks
Data frame of blank data used for QA, or NULL if not provided.
- population
Named list or NULL. Per-plate population/noise parameters for per-plate arms (single-plate Bayes: this plate's
sigma_obs/nuand hyperprior-informedmu_*; frequentist:sigma_resid). Same shape as the multiplatepopulationslot (seenew_calibration_result_multiplate()). NULL for the pooled multiplate arm, where group params live on the multiplate container instead.