NEWS.md
B in (n-out-of-n) empirical / multiplier / subsampling bootstrap samples on coverage when compared to sandwichSignificance heading in our summary output to be abbreviated to Signif: to be consistent with lm() output2/3 digitsR/scripts_and_filters/experiments/ dir, remove old experimentsmaars function metadata from pkgdown
base:: prefix use e.g. base::return()
dplyr::summarise to dplyr::summarize for spelling consistencystats functions use the stats:: prefix.data related rlang issuesdf since this conflicts with stats::df. We should change this after the demoBoston Housing Dataset with citations. Add unit tests for the correctionsswitch based approachmodel.matrix in sandwich variance and use residuals in the computationmaars to have a package level doc. Add @importFrom statementsNOTE by adding .gitkeep in vignettes to .RBuildignore
NOTE by DESCRIPTION meta-information by making it a couple of sentences. This is a placeholder and we should refine it before official CRAN release.boston-housing.R and la-county.R files into a single data-maars.R files. Consolidate test files accordinglystyling to our code using the Makefile and styler::style_dir(here::here('R')) and for tests
make style which does both R and tests
url_check() are all resolved for CRAN
inst/REFERENCES.bib since they can cause CRAN url issuespkgdown our site.%>% and base code, and just break pipes into variablesattr(obj, "class") <- c("obj_class_name") to be of the form class(obj) <- "obj_class_name" for consistencydplyr::_all scoped words using the superseded across functiondigits = 3 formatting in summary.maars_lm printed outputsummary.maars_lm, print.maars_lm, plot.maars_lm methods into maars-lm.R filelm-var.R code into into sandwich-var.R since they are both default estimatorsla_county tibble directly from maars in our vignetteget_summary() functionget_plot(), get_summary(), summary.maars_lm(), and get_confint(), confint.maars_lm() functionality for maars_lm() objectsplots function with default options and wrap around get_plot()
lm() outputmaars_lm plot outputmaars standard errors. So sand, well_specified assumptions are printed by default, and other standard errorscomp_lm_var using glue::glue() and improve formattingget_confint() to return tidy tibble outputterm, conf.low, conf.high for confint(maars_lm). Fix it to be the same as confint(lm)
la_county tibble to our packageREADME.Rmd to use devtools rather than remotes as the preferred package installationdevtools to Suggests in DESCRIPTION so that vignettes use our latest codela_county tibble is documented methodlm() modelweights type in assumptions for multiplier bootstrapn in assumptions for empirical bootstrapmaars_lm object from comp_var
as.maars_lm function to be run on an lm objectconfint.maars_lm methodprint.maars_lm methodsummary.maars_lm methodplot.maars_lm, currently for lm objectsmaars release 0.6.0 in NEWS.md
GPL-2 and GPL-3 LICENCE
MIT LICENCE filescomp_var and of the estimators of the variancecomp_var
get_assumptions)V of coefficients estimates for the bootstrapweights_type for comp_boot_mul to be default rademacher
comp_var to handle the list outputs generated by the estimators of the varianceget_summary within the functionget_summary
comp_var
NULL for empirical bootstrap, multiplier bootstrap, residual bootstrapcomp_var
if-then-else skeleton for comp_var
residual bootstrap to comp_var
comp_var for empirical bootstrap, multiplier bootstrap, residual bootstrapqr from empirical bootstrap namespkgdown references into suitable categories like usethis, here is an example
#' @export in roxygen2 and replacing it with #' @keywords internal
.gitattributes fileset.seed in all our vignette chunks. This is for reproducibility, but to also avoid merge conflicts every time we rebuild the package using make build_package
@details responsibly. See below
var_name with \code{var_name}. See below
1.96 with appropriate outputs from statistics functions, in vignettes
ggplot2 code into separate plots and then combine them using +. This will make it much easier to manage the code and make it readableggplot2 themes used in our plots. See ggplot2 theme wrapper below
ggplot2 font and other settings used in our plots. As a preference we should only use labs for example. See ggplot2 theme wrapper below
names_prefix = "q" to pivot_wider for the ggplot2 code i.e. to avoid .data$0.275 issue. Note the "q" here stands for quantile.utils-common.R similar to the selectiveInference package, ensure that common functions are put in this filetests directory i.e. delete unused R test filesR directory i.e. delete unused R files#' @importFrom rlang .data in all our functions under #' @export
R CMD CHECK error caused in vignette, namely this one
stringr to DESCRIPTION
multiplier bootstrap weights to be applied inside purrr e.g using purrr::map2 rather than generating the e matrix up front. Conduct benchmarks of this to test speed, so that current version still has it’s place in benchmarkingvignette should knitr::kable package instead of DT package and just need to ensure Table 1 is correctly populatedvignette we just need to ensure Table 1 is correctly formatted i.e. use LaTeX for headings, consistent width, and captions. See here for the use of LaTeX in knitr::kable heading rowsRmd line wrap hooks similar to Yihui Xie’s approach and the oxforddown approach
weight_type in multiplier bootstrapB in multiplier variancestd_gaussian, rademacher, and mammen, webb, gamma
Licence and License.md
multiplier bootstrap for lm standard errorsempirical bootstrap for lm standard errorsempirical bootstrap for glm standard errorslm_fit to be mod_object in all of our functionsstats::, dplyr::, etct-test values in Table 1 of the Models As Approximations - Part 1 paperB must be a positive integerReadme.Rmd to add installation instructions and link to official package