The R code that makes the study's ARD

One cards / cardx call per analysis row, each result under a name of its own (ard_<analysis_id>) and tagged with its output_id, analysis_id and population_id (tag_ard()), bound into one ARD and saved to the study key output (default output/ard/ard.rds). The code runs from the study folder, and reads as a person writes it: the tidyverse layout (a call on one line when it fits in 80 characters, else an argument a line), dplyr verbs for the data (filter(), mutate(), select()), and a custom analysis's code with the program's names in place of data and population (in local() only when it needs a scope of its own: it defines a function, or assigns a name the program has).

Usage

tfl_ard_code(
  spec,
  output_id = NULL,
  save = TRUE,
  part = c("all", "setup", "body"),
  statistics = NULL,
  methods = NULL,
  dir = ".",
  codelists = NULL
)

Arguments

spec

An tfl_ard_spec() (or the path of one), or a study of the YAML spec (tfl_read_study_spec()): its data catalog and its reports' ard.analysis_data / ard.analyses, the reports in the TOC's order, its code lists the reports' own (when codelists is NULL).

output_id

Only these outputs' analyses; NULL for all.

save

FALSE leaves out the final saveRDS(). For one report's "body", TRUE ends it in save_ard() (its rows into the study ARD, with the definition's fingerprint), FALSE in ard.

part

"all" (the whole program), "setup" or "body".

statistics, methods

The catalogs (tfl_ard_statistics(), tfl_ard_methods()) to use instead of the current ones.

dir

The study folder: the fingerprints saved with the ARD (tfl_ard_spec_hash()) read the study's own function files from it.

codelists

The reports' code lists: a table definition (its codelists sheet) or a data frame with output_id, variable, value, label and order. A code list is a report's: every row names its report (a blank output_id stops). A report's rows of the variables its analyses read (by, strata, variables, and the names in args, code and post) count: the program writes them at its head (cl_sex <- c(F = "Female", M = "Male")) and puts them on the data the analyses read (set_levels(SEX = cl_sex)): each column a factor in the list's order, its values the labels (a value without one stays itself) – the ARD holds "Female", and counts a level no record has (n = 0). A value the list does not have (not NA) stops the program. An analysis's where on such a column is written in its labels: one that names a value whose label differs stops here. In the study's program, each report's part reads its data again with its own. NULL (default): the data as read.

Value

The code, one element per line.

Details

The formats (the method's and the row's formats, over each statistic's default in tfl_ard_statistics()) are in the cards call itself, its fmt_fun argument, where a reader sees them next to the statistics: fmt_fun = everything() ~ modifyList(fmt_default, list(mean = 2L)), a variable's own (BMIBL:sd=3) with its own list. An integer is that many decimals, xx.x% is label_round(1, scale = 100) and pvalue the program's fmt_pvalue() (<0.001 or 3 decimals); cards::apply_fmt_fun() then fills stat_fmt. What takes no fmt_fun gets the same formats after the call, from fmt_ard(): cardx's tests, CIs and models, cards::ard_stack_hierarchical(), a study's own function, custom code, ard_pairwise(), ard_stack()'s own rows (the by counts, the total N), an analysis whose args gives fmt_fun or whose post changes the ARD, a variable's own format for ard_hierarchical() or ard_tabulate_rows(). stat_fmt is the same either way.

The functions the program calls – set_levels(), tag_ard(), fmt_ard(), keep_stats(), fmt_pvalue(), save_ard() – are tfl_helpers_code()'s: the whole program (part = "all") carries them, and a study keeps them in a file of its own, so its programs run without tflspec.

part gives a piece of it instead, for a program layout of one's own (tflplanner writes one program per output that sources a shared setup): "setup" is what every piece starts with – library(cards), library(dplyr), the computed statistics the analyses use (tfl_stats) and each statistic's format (fmt_default); "body" is the analyses of output_id, starting with report_id <- "..." and its code lists (cl_<variable>), and ending in save_ard() (one report, save = TRUE) or ard.