Parameterized RMarkdown with non-value targets

[arnold-c]

I'm not sure if this is outside of the realm of targets and archetypes, but I was wondering if there's a better way to use parameterized reports that rely on non-value targets? What I mean by that is that I have different data subsets and model fits that follow a naming pattern, and I want to use one set in each report. What I've come up with is a bit of a hack and has the issue that it doesn't track the dependencies so I need to manually make and invalidate targets. I could explicitly specify all the targets using tar_load(), but my hesitancy there is that there would be a lot of unnecessary targets loaded in each branch, and I think I'd still need to do something like below in order to select the correct targets to use in the report.

lines <- c(
  "---",
  "title: report",
  "output_format: html_document",
  "params:",
  "  cyls: 4",
  "---",
  "",
  "```{r}",
  "data <- sym(glue::glue('data_{params$cyls}_cyls'))",
  "```",
  "",
  "```{r}",
  "tar_load(all_of(data))",
  "```",
  "",
  "```{r}",
  "data <- eval(data)",
  "```",
  "",
  "```{r}",
  "data %>% summarize(mean(disp), mean(drat))",
  "```"
)
writeLines(lines, "report.Rmd")

library(targets)
tar_script({
  library(tarchetypes)
  library(tidyverse)
  options(crayon.enabled = FALSE)
  list(
    tar_target(data_4_cyls, filter(mtcars, cyl == 4)),
    tar_target(data_6_cyls, filter(mtcars, cyl == 6)),
    tar_render_rep(
      report,
      "report.Rmd",
      params = data.frame(
        cyls = c(4, 6),
        output_file = c("f1.html", "f2.html")
      )
    )
  )
})
tar_make()

[wlandau]

I think you will have more luck with tar_eval(): https://books.ropensci.org/targets/static.html#metaprogramming

file.create("report.Rmd") # Does not require tar_read() or tar_load() if the data gets sent through Rmd parameters.
#> [1] TRUE

library(targets)
tar_script({
  library(rlang)
  library(tarchetypes)
  tar_eval(
    list(
      tar_target(data, filter(mtcars, cyl == x)),
      tar_render(report, "report.Rmd", params = list(data = data))
    ),
    values = list(
      x = c(4, 6),
      data = syms(c("data_4", "data_6")),
      report = syms(c("report_4", "report_6"))
    )
  )
})

tar_manifest(fields = command)
#> # A tibble: 4 × 2
#>   name     command                                                              
#>   <chr>    <chr>                                                                
#> 1 data_4   "filter(mtcars, cyl == 4)"                                           
#> 2 data_6   "filter(mtcars, cyl == 6)"                                           
#> 3 report_4 "tarchetypes::tar_render_run(path = \"report.Rmd\", args = list(para…
#> 4 report_6 "tarchetypes::tar_render_run(path = \"report.Rmd\", args = list(para…

tar_visnetwork()

^{Created on 2022-03-24 by the reprex package (v2.0.1)}

[arnold-c]

Thanks for the quick response. I'll give it a try and get back to you.