Building reproducible workflows with {targets}

Author

Affiliation

Andrew Heiss

Georgia State University

Overview

Have you ever created files named “01_clean.R”, “02_analysis.R”, and “03_plots.R” and forgotten to run one of the files in the right order? Have you had to manually rerun an entire analysis after a single input changes—even if that input only affected one plot? Data analysis involves many moving parts. Traditional script-based approaches often lead to fragile pipelines, wasted computation, and reproducibility nightmares. Fortunately, the {targets} package for R offers a powerful solution by implementing a Make-like pipeline toolkit that transforms chaotic scripts into streamlined, automated workflows that intelligently skip unnecessary steps.

Materials

All the materials we’ll use today are three self-contained RStudio projects:

 targets-penguins.zip    08-comparisons-original.zip    08-comparisons-targets.zip

Packages

You’ll need all these packages on your computer if you’re following along:

install.packages(
  c("tidyverse", "ggrepel", "sf", "tigris", "dataverse", "jsonlite", 
    "targets", "tarchetypes", "tidybayes", "brms",
    "marginaleffects", "modelsummary", "parameters", "patchwork"
  )
)

Slides

 Open slides in new window

Other examples and resources

Official documentation

• {targets} book
• {targets} site
• {tarchetypes} site (extra functions for rendering to Quarto or R Markdown, using a URL as a storage format, conditionally controlling target execution, etc.)

Pipelines for research projects

• “Navigating Hostility: The Effect of Nonprofit Transparency and Accountability on Donor Preferences in the Face of Shrinking Civic Space”
  Analysis notebook + preprint and _targets.R

• “Derogations, Democratic Backsliding, and International Human Rights During the COVID-19 Pandemic”
  Analysis notebook + preprint and _targets.R

• “Pandemic Pass? Treaty Derogations and Human Rights Practices During COVID-19”
  Analysis notebook + preprint and _targets.R

Pipelines for course websites

• Data Visualization with R
  (pipeline builds data, zips project folders, converts HTML slides to PDF, builds Quarto website, and uploads everything to a server)
  Website and _targets.R

• Program Evaluation for Public Service
  (pipeline builds data, zips project folders, converts HTML slides to PDF, builds Quarto website, and uploads everything to a server)
  Website and _targets.R

• Governance Around the World
  (pipeline builds Quarto website and uploads everything to a server)
  Website and _targets.R

Pipelines for Idaho’s 2024 election

• DIY Election Desk (example repository with two independent pipelines: one for processing data and one for building a site)
• “Election Night Reporting Using R & Quarto” posit::conf(2025) talk