NEWS

simhelpers 0.3.1.9999

Added option to specify a unique ID variable for every repetition in repeat_and_stack() and bundle_sim().
Refactored bundle_sim() so that it uses repeat_and_stack() internally.
Refined evaluate_by_row():
- Refactored so that it uses only variables that match argument names of the function to be evaluated.
- Added an option nest_results to return a nested column of results (for more compact storage).
extrapolate_coverage() and extrapolate_rejection() gain an exclude_above argument to exclude results based on more than a certain number of bootstraps from being included in the extrapolation calculations.

Added support for bias-corrected and bias-corrected-and-accelerated (BCa) bootstrap confidence intervals.
Corrected an error in the documentation of bundle_sim().

Added functions for calculating bootstrap p-values and confidence intervals and for estimating rejection rates, coverage rates, and interval widths by extrapolating across bootstrap subsamples.
Added repeat_and_stack() function, which is similar to the base R replicate() and to the now deprecated purrr::rerun(), with the option to stack the output into a single data.frame.
Added "stddev" as a performance criterion in calc_absolute().
Added winsorize options in calc_absolute(), calc_relative(), calc_relative_var(), and calc_coverage().

Added new, experimental function bundle_sim() to compose a set of functions into a simulation driver.
Added an argument to evaluate_by_row() to control the name of the variable where simulation results are stored.
Revised the calc_*() functions so that they can take vectors or variable names from a specified dataset.
calc_rejection() can now compute rejection rates for multiple alpha levels.
Renamed the K variable computed in the calc_*() functions to avoid over-writing variables when using multiple performance calculations inside of dplyr::summarize().
Updated syntax in vignettes and examples to use current tidyverse conventions.