Skip to content

A Workflow for R

Metadata

  • Author: David Smith
  • Full Title: A Workflow for R
  • Category: #Type/Highlight/Article
  • URL: https://blog.revolutionanalytics.com/2010/10/a-workflow-for-r.html

Highlights

  • Transparency: A good workflow organizes the elements of the project logically and clearly, to make it easy for an observer (including yourself) to understand how the pieces come together.
  • Maintainability: A good workflow makes it easy to modify and adapt the project. Standardized script names and good commenting practices (in the code, as well as things like README files) are key here.
  • Modularity: A good workflow encapsulates discrete tasks into separate components (e.g. scripts), so that it’s always clear where modifications need to be made (and only made in one place), and components are re-usable for other projects.
  • Portability: A good workflow makes it easy to move the project to another system, or hand it over to another person to work on, in such a way that it can still easily be run elsewhere. (By using relative (not absolute) pathnames, and remote access to sharedWorkflow for statistical analysis and report writing data, are two examples.)
  • Tags: favorite
  • Reproducibility: A good workflow makes it easy for you, or others, to reproduce your results.
  • Efficiency: Here I’m referring to the efficiency of you, the programmer, not computational efficiency. A good workflow saves you time, by making it easier to work on the project, and by automating as much of the process as possible.