Reproducible Analytical Pipelines

My personal hot takes 🔥

Dr Rhian Davies

2026-05-21

🏛️ My RAP Pillars

• 🔍 Visible: Can I find it?

• 🔁 Reusable: Can I adapt it?

• 🛡️ Reliable: Can I trust it?

🔍 Visible

• Public and open 🔓
• Discoverable 🌐
• Clear README & examples 📖
• LICENSE included 📃

🔁 Reusable

• Open-source languages 🔓
• Well documented 📚
• Consistent code style 💅
• Modular 🧩
• Configurable ⚙️
• Easy to install 📦

🛡️ Reliable

• Tested 🧪
• Automated checks 🤖
• Peer reviewed 👩‍💻
• Reproducible environments 🌱

Commented code

\(\neq\) good code

• ❌ Over-commented
• ✅ Self-documenting

library(dplyr) # load dplyr package
d1 <- as.data.frame(crimtab) # turn table into dataframe
d2 <- d1 |> # create new dataset
  mutate(# make new variables
    fg = f1(Var1), # convert Var1 into numeric this represents finger length
    ht = f1(Var2),     # convert Var2 into numeric this represents height
    n = Freq # rename frequency column this is the count
  ) |>
  select(fg, ht, n) # keep only required columns

# function to turn factor into numeric 
# this is needed because the variables are factors
# and we want numeric values instead
f1 <- function(x) {
  y <- as.character(x)  # convert factor to character
  as.numeric(y)  # convert character to numeric
}

library(dplyr)

criminal_summary <- crimtab |>
  as.data.frame() |>
  mutate(
    finger_length = extract_numeric_from_factor(Var1),
    height = extract_numeric_from_factor(Var2),
    count = Freq
  ) |>
  select(finger_length, height, count)

extract_numeric_from_factor <- function(factor) {
  as.numeric(as.character(factor))
}

Reusable ♻️

not reproducible

Open first 🔓

You don’t need friends 👩‍💻

to use Git

but it helps

Anyone

can build an R package 📦

Myths ❌

It doesn’t have to…

• have hundreds of functions
• be useful for anyone else
• do anything fancy
• be on CRAN
• be scary

Ditch QA

(spreadsheets). Make QA work for you.

It doesn’t have to be

perfect ✨

How do I get support? 💙

• RAP drop-in sign up & discussions
• RAP Code Review Buddies Network
• NHS Open Analytics slack #rap
• Chat with me @statsRhian 😊