Skip to contents

report and performance support for bmmfit objects

Source: R/easystats-report.R
easystats-report-bmmfit.Rd

These S3 methods let report::report() describe a bmm fit as the model you specified with bmm(), and let performance::model_performance() return only the indices that are defined for bmm models. They build on the insight methods described in easystats-bmmfit.

Usage

# S3 method for class 'bmmfit'
format_model(x)

# S3 method for class 'bmmfit'
report_model(x, table = NULL, ...)

# S3 method for class 'bmmfit'
report_random(x, ...)

# S3 method for class 'bmmfit'
report_priors(x, ...)

# S3 method for class 'bmmfit'
report_info(x, parameters = NULL, ...)

# S3 method for class 'bmmfit'
model_performance(model, metrics = c("LOOIC", "WAIC"), verbose = TRUE, ...)

Arguments

x, model

A bmmfit object (created by bmm()).

table

Ignored, for consistency with the generic.

...

Passed on to the methods of report and performance for brms fits.

parameters

The result of report::report_parameters() for x, or NULL to compute it.

metrics

The indices to return: "LOOIC", "WAIC", or "all" and "common" for both.

verbose

Should a message name the requested indices that are left out?

Value

See the generics, e.g. report::report_model() and performance::model_performance().

Details

  • format_model() names the model and its version, e.g. "Bayesian Signal Discrimination Model (SDM) (simple)".

  • report_model() describes the model in one sentence, with all response variables of models such as ezdm() or m3() in one list, the formula of each model parameter as summary() prints it, and the grouping factors of the random effects.

  • report_priors() gives the text of prior_info(fit, format = "text"): the priors in the names of the model parameters, with a note that they apply on the sampling (link) scale. The short version of the report, summary(report(fit)), leaves the priors out.

  • report_info() keeps the paragraph of report that explains the parameter summaries, and replaces its sentence on convergence (the thresholds for R-hat and ESS) with the sentence of report_convergence(fit, format = "text"): the largest R-hat, the smallest bulk and tail ESS and the number of divergent transitions, without verdicts.

  • model_performance() returns LOOIC and WAIC (with ELPD and the standard errors). R2, RMSE and the other indices of performance need the expected response or residuals, which bmm models do not define; requested ones are left out with a message, and when none of the requested indices is defined, LOOIC and WAIC are returned.

The parameter summaries, with their R-hat and ESS, come from report unchanged. report lists no parameter estimates for a model without predictors, such as an mpt() or ezdm() fit with intercepts only; use summary() for those. report_methods() drafts a methods paragraph from the same facts in bmm's own words.

See also

prior_info(), report_methods(), report_convergence(), easystats-bmmfit

Examples

if (FALSE) { # \dontrun{
report::report(fit)
performance::model_performance(fit)
} # }