check_results: Check results from MPTmultiverse
In MPTmultiverse: Multiverse Analysis of Multinomial Processing Tree Models

Description Usage Arguments Details Examples

View source: R/check_results.R

Set of helper functions that allow checking if model estimation worked as intended. Depending on the method and function, these functions return slightly different information.

check_results(results)

write_check_results(DATA_FILE, results, append = FALSE)

check_set(results)

`results`	An object of class multiverseMPT.
`DATA_FILE`	character string. File name to use.
`append`	logical. If `TRUE`, output will be appended to `DATA_FILE`; otherwise, it will overwrite the contents of `DATA_FILE`.

check_results prints relatively verbose output detailing diagnostic information for each method to the console. For the frequentist methods, this is based on either the rank of the observed Fischer (or Hessian) matrix of the MLE estimate or based on empirical identifiability (based either on repeated re-runs or the width of the bootstrapped parameter distribution). For the Bayesian methods, this is convergence statistics R-hat and number of effective samples. write_check_results writes the results of check_results to a specififed file (instead of printing it to the console).

check_set returns a tibble with one row, where each expected method corresponds to a column with a boolean (TRUE/FALSE) value. Entries TRUE correspond to no problem and FALSE correspond to problems. FALSE means the method is either missing from the results file or (for the Bayesian methods) there are core parameters for which the convergence criteria defined in getOption("MPTmultiverse") are not met.

load(file = system.file("extdata", "results_bayen_kuhlmann.RData",
                        package = "MPTmultiverse"))
## prints checks to console
check_results(results)

## returns tibble with single row
check_set(results)