Everything you wanted to know about R & Quarto

But were too afraid to ask

Rhian Davies

Oct 17, 2024

Hi I’m Rhian 👋

• Senior Data Scientist at the Strategy Unit
• Expert in R training & consultancy
• Statistics PhD
• Statistical Ambassador for the RSS

• Evaluation and impact assessment

• Complex modelling

• Evidence reviews

• Learning & Development

• HACA and NHSR conference

What & Why R?

What questions do you have? 🤔

• Ask a question
• View all questions

What is R?

• A programming language
• About 30 years old
• Originally specialised in statistics but now very broad
• Cross-platform
• Used by many organisations

Silly R stuff too?

Why use R?

• Free & open-source
• Diverse ecosystem of packages
• Write your own functions
• Excellent support for statistical modelling
• Professional graphics
• Students learn R
• Friendly community 💜

Many of the statistical models were written by the statisticans who developed the methods

Reproducible Analytical Pipelines ⭐

It will be built using open source software for data management, analysis and visualisation (such as R or python) as this is standard, portable, and available to all for checking and re-use.

— Goldacre Review 2022

Packages 📦

• Over 21,000 on CRAN
• Many more on GitHub
• Diverse topics

Can it work with my data?

• Yes!

SQL databricks databases SPSS SAS CSV / Excel Cloud e.g. Azure API

How do I get started? 🚀

1. Install R

2. Install RStudio

You can also try it out in Posit Cloud

Demo R in terminal vs RStudio You don’t need to install R for today’s session But I do have some snippets of code you can play with

If you’re already a coder, there’s a new IDE called Positron based on VS-Code which you might prefer.

Let’s go! 🚀

Length of Stay data

Fictional data, adapted from {NHSRdatasets} We might want to ask questions such as - Which area has the most patients? - What is the longest length of stay in the data? - Is there a correlation between age and length of stay?

Load in data

Watch me 👀

Your turn 💻

Change the code below and click Run

Important

Capitalisation matters!

Watch me 👀

• Keep just the Midsomer data

Note

Notice double ==

Your turn 💻

• Change to keep Sex is Female

Watch me 👀

• What is the mean length of stay for each sex?

Your turn 💻

• Calculate the median() LOS for each Area

Note

Feel free to try grouping by other variables, or using other summarisation functions e.g. mean(), min(), max(), sd()

Making pretty graphics

Professional quality graphics

Professional quality graphics

NEW on coronavirus: many western countries may soon face Italy’s situation

Case numbers since outbreaks began in several countries have tracked a ~33% daily rise. This is as true for UK, France, Germany as Italy; the latter is simply further down the path https://t.co/VcSZISFxzF pic.twitter.com/xM6wXuMk4n

— John Burn-Murdoch (@jburnmurdoch) March 11, 2020

Tweet from jburnmurdoch

Watch me 👀

Your turn 💻

1. Change the the colour of the points to "blue"
2. Add the line geom_smooth(method = "lm"). You’ll need a + at the end of line 4.

We can make maps too

No need for expensive GIS software

Shiny apps

• Interactive web applications
• No need to learn Javascript or HTML
• Gallery
• Mental Health Surge Modelling

Hello quarto 👋

What is quarto?

An open-source scientific and technical publishing system

Artwork from “Hello, Quarto” keynote by Julia Lowndes and Mine Çetinkaya-Rundel, presented at RStudio Conference 2022. Illustrated by Allison Horst.

These slides were made in Quarto!

Automating reports 📋

• The data was wrong - can you re-run the analysis?
• Your manager loves it - can you do it for another region?
• Next month - do it all again.
• There must be a better way

Example report

statsrhian.github.io/example-quarto-report/example-report.html

Template based-reports

---
title: "NHS Workforce Statistics for `r params$ics_name`"
subtitle: "Data for `r params$month_year`"
author: "Maria Garcia"
date: "2023-09-28"
params:
  ics_name: "North East and North Cumbria"
  month_year: "April 2023"
---

Interweave text and code 🧶

Read in the data

filename = glue("NHS Workforce Statistics, {params$month_year} England and Organisation.xlsx")

Clean the data

staff_group = 
  staff_group |>
  filter(`ICS name` == params$ics_name) |>
  select(`Organisation name`, `Total`,
         `HCHS Doctors`, `Nurses & health visitors`,
         `Midwives`, `Ambulance staff`) 

Add insight

The table below shows the total number of doctors and nurses for each organisation within `r params$ics_name`. We can see that the organisation with the most midwives is the `r pull(max_midwives, "Organisation name")` with `r round(max_midwives$Midwives)` staff.

Re-run next month

---
title: "NHS Workforce Statistics for `r params$ics_name`"
subtitle: "Data for `r params$month_year`"
author: "Maria Garcia"
date: "2023-09-28"
params:
  ics_name: "North East and North Cumbria"
  month_year: "May 2023"

Report a different region

---
title: "NHS Workforce Statistics for `r params$ics_name`"
subtitle: "Data for `r params$month_year`"
author: "Maria Garcia"
date: "2023-09-28"
params:
  ics_name: "Lancashire and South Cumbria"
  month_year: "May 2023"

Using R in your organisation

Enterprise tools 🏢

• Posit Workbench
• Posit Package Manager
• Posit Connect

Working with Posit

Posit Partners

Is R safe? ✅

• All code is open
• Packages are often well tested
• R Validation Hub Posit Partners

Want to learn more? 📚

• R4DS book
• TidyTuesday
• The NHS-R community