Don't repeat yourself, talk to yourself! Repeated reporting in the R universe

bit.ly/sharstudioconf

@SharlaGelfand

The reports took all year!

They weren't reproducible

They were scattered,

hard to keep track of, boring, manual, repetitive

not easily repeatable

Reproducibility!

Code all in one place!

It'll be worth it!

We can do new things!

Next year arrives

SPSS, Excel, Word

Not reproducible

Difficult to keep track of

Manual and repetitive

Not easily repeatable

Time consuming

SPSS, Excel, Word

Not reproducible

Difficult to keep track of

Manual and repetitive

Not easily repeatable

Time consuming

My R code

Kind of reproducible

Difficult to keep track of

Manual and repetitive

Not easily repeatable

Time consuming

Don't repeat yourselfTalk to yourself!14

Don't repeat yourself

Abstract away repetitive logic and automate repetitive processes

Copy and pasting is error prone

Focus should be on what changes

Copying over and over and redoing work is pointless

Talk to yourself

It's easier to document why you're doing something now than to remember why you did it a year ago

Focus on future-self instead of present-self

Easy-to-read code doesn't explain motivations

I cleaned up after myself

Reorganize files

Default ordering

Setup file

I made functions

Body: what stays the same

Argument: what changes

Chatty functions

library(usethis)
ui_info("hi!")

Chatty functions

analysis_year <- 2019
ui_todo("Querying demographics data for
        {analysis_year}...")

Chatty functions

I made code templates

Artisanal data doesn't require artisanal code

Used to focus on cleaning this data really well

Too specific!

No idea why I did something

Code templates

Generalizable code

Super specific instructions

Deal with artisanal data instead of remembering what to do

Prescriptive templates

Automating templates

clean_data <- function(analysis_year) {
  usethis::use_template(
    template = "clean_data.R",
    save_as = "01_clean-data.R",
    data = list(analysis_year = analysis_year),
    package = "demographicsreport"
  )
}

Automating templates

clean_data <- function(analysis_year) {
  usethis::use_template(
    template = "clean_data.R",
    save_as = "01_clean-data.R",
    data = list(analysis_year = analysis_year),
    package = "demographicsreport"
  )
}

Automating templates

clean_data <- function(analysis_year) {
  usethis::use_template(
    template = "clean_data.R",
    save_as = "01_clean-data.R",
    data = list(analysis_year = analysis_year),
    package = "demographicsreport"
  )
}

I put it all together in a package

Why make a package?

Organize functions

Home for templates

Better documentation

Share code with coworkers