Summarize the parameter estimates, run details, and any heuristics of a bootstrap run, saving the results to a boot_summary.RDS data file within the bootstrap run directory.

summarize_bootstrap_run(.boot_run, force_resummarize = FALSE)

bootstrap_estimates(.boot_run, format_long = FALSE, force_resummarize = FALSE)

get_boot_models(.boot_run)

Arguments

.boot_run

A bbi_nmboot_model object.

force_resummarize

Logical (T/F). If TRUE, force re-summarization. Will only update the saved out RDS file when specified via summarize_bootstrap_run(). See details for more information.

format_long

Logical (T/F). If TRUE, format data as a long table, making the data more portable for plotting.

Details

  • bootstrap_estimates() quickly extracts and formats the parameter estimates from each model run. If the data was previously summarized, the data will be read in instead of re-executing (this can be overridden via force_resummarize = TRUE).

  • summarize_bootstrap_run() does the following things:

    • Tabulates run details and heuristics.

    • Calls summary_log() and binds the results to the parameter estimates.

      • bootstrap_estimates() will include this appended model summary information if a boot_summary.RDS data file exists.

    • Either saves this data out to boot_summary.RDS, or reads it in if it already exists (see section below).

    • Formats the returned object as a bbi_nmboot_summary S3 object, and displays key summary information when printed to the console.

Saved out data file:

The first time summarize_bootstrap_run() is called (or if force_resummarize = TRUE), it will save the results to a boot_summary.RDS data file within the bootstrap run directory. If one already exists, that data set will be read in by default instead of being re-summarized.

  • The purpose of this is functionality two fold. For one, it helps avoid the need of re-executing model_summary() calls for a large number of runs. It also helps to reduce the number of files you need to commit via version control (see cleanup_bootstrap_run()).

Functions

  • summarize_bootstrap_run(): Summarize a bootstrap run and store results

  • bootstrap_estimates(): Tabulate parameter estimates for each model submission in a bootstrap run

  • get_boot_models(): Read in all bootstrap run model objects

Examples

if (FALSE) {

.boot_run <- read_model(file.path(MODEL_DIR, "1-boot"))
boot_sum <- summarize_bootstrap_run(.boot_run)

# Optionally compare to original estimates
param_estimates_compare(boot_sum)


# Long format is helpful for plotting estimates:
bootstrap_estimates(.boot_run, format_long = TRUE) %>%
dplyr::filter(grepl("THETA", parameter_names)) %>%
  ggplot(aes(x = estimate)) +
  facet_wrap(~parameter_names, scales = "free") +
  geom_histogram(color = "white", alpha = 0.7) +
  theme_bw()

}