Calibration result objects

Objects of class calibrar.results are returned by calibrate(). They store the final parameter values together with the objective value, convergence information, the set of calibrated parameters, and tracing information from the calibration run.

coef() returns the fitted parameter values stored in a calibrar.results object.

plot() is currently a placeholder method for calibrar.results objects. No graphical summary is produced at present.

dim() returns the number of fitted parameters after unlisting the parameter structure stored in the result object.

Usage

# S3 method for class 'calibrar.results'
print(x, ...)

# S3 method for class 'calibrar.results'
coef(object, ...)

# S3 method for class 'calibrar.results'
predict(object, replicates = 1, na.rm = FALSE, ...)

# S3 method for class 'calibrar.results'
plot(x, ...)

# S3 method for class 'calibrar.results'
summary(object, ..., show_par = NULL, par.only = FALSE)

# S3 method for class 'summary.calibrar.results'
print(x, digits = 3, ...)

# S3 method for class 'calibrar.results'
dim(x)

Format

A list, usually with the following components:

par: Best parameter values found, returned with the same structure as the input par supplied to calibrate().
value: Objective-function value at par.
counts: Named integer vector giving the number of function and gradient evaluations recorded for the last calibration phase.
convergence: Integer convergence code returned by the optimiser used in the final phase.
message: Character string with additional information returned by the optimiser.
method: Name of the optimisation method used in the calibration.
fn: Objective function used for the calibration.
active: Logical vector indicating which parameters were active (estimated) in the full calibration problem.
elapsed: Elapsed time for the complete calibration run.
trace: A list containing tracing information from the last phase and additional run-level metadata. Depending on the control settings and the optimiser used, this may include per-phase outputs, generation traces, partial objective values, and other diagnostic information.

Arguments

x: A calibrar.results object, or a summary.calibrar.results object for print.summary.calibrar.results().
...: Additional arguments passed to the underlying method.
object: A calibrar.results object.
replicates: Number of replicate evaluations of the fitted parameter set to perform. If replicates = 1, the underlying model function is evaluated once. If replicates > 1, repeated evaluations are combined by taking the mean across replicates for each returned output.
na.rm: Logical. Passed to the internal averaging step used when replicates > 1.
show_par: Optional character vector of parameter names or integer vector of positions indicating which parameters should be shown in the summary. By default, all fitted parameters are shown.
par.only: Logical. If TRUE, return only the method name and selected parameter values. This is mainly useful when summarising several calibrar.results objects side by side.
digits: Number of significant digits to use when printing summary tables or parameter values.

Value

coef.calibrar.results() returns the fitted parameter values, with the same structure as the par component of the object.

predict.calibrar.results() returns the model outputs produced by the fitted parameter set. When replicates > 1, it returns the mean output across replicates and stores the full replicate-level outputs in the .replicates component of the returned object.

plot.calibrar.results() returns NULL, invisibly.

summary.calibrar.results() returns either:

an object of class summary.calibrar.results when called on a single object; or
a data.frame with one row per result object when several calibrar.results objects are supplied through ....

The print method formats these summaries for console display.

dim.calibrar.results() returns a single integer.

Details

A calibrar.results object represents the final output of a complete calibration workflow, potentially including multiple calibration phases.

The most useful accessors are:

print() for a compact console summary;
summary() for a more explicit summary, including selected parameters;
coef() to extract the fitted parameter values;
predict() to evaluate the fitted parameter set with the underlying model function when this is available; and
dim() to report the number of parameters in the fitted solution.

The predict() method is currently only available when the objective function was built with calibration_objFn(), because it relies on metadata attached to that function.

Usage

Format

Arguments

Value

Details

See also