Slides for the presentation Caring About Code. This talk was initially presented at the Plant Production System seminar at Wageningen University. It is mostly aimed at R beginners.

Link to the slides: https://alanguillaume.github.io/CaringAboutCode/

Code should be understood as a research project blue-print that allows one’s results to be easily reproduced by anyone starting from their author. However, code should not only be machine-readable but also human-readable and, as such, should be treated as a communication medium in its own right. If tackling your coding problems with that given mindset, you can directly use what you already know about writing in general to improve your code. That boils down to:

1. Enough vocabulary: to simply know what to google.

   • The Basics: SWIRL.
   • To get stuff done: the tidyverse ; R for Data Science.
   • To become an R ninja: Learn a bit more about R inner-clockworks, Advanced R (chapters 3 and 9, to start with).

2. Clear Syntax: Once you know enough words you can arrange them in properly layed-out sentences.

   • Let the code breathe: Code also likes spaces and escaped lines.
   • Be explicit: in the way you name things.
   • Be consistent: in the way you name things (bis) but…
   • DRY: Don’t Repeat Yourself.
   • The Torah Syndrome: limit yourself to 80 characters per line.

3. Clear Structure: Once you know how to write sentences you can arrange them in paragraphs. The general idea of one paragraph = one idea also holds for code and comes in the name of modularity.

   • Write a few small scripts each doing one thing well.
   • Whenever needed, write functions (automation recipes) and put them all together in a “toolbox” script.

All the above points are illustrated in a companion project: blackadder. In that small example, I made a heavy use of the tidyverse especially stringr and purrr to clean data (Enough Vocabulary). The data and analysis are a bit unorthodox but the objective was to focus on the code layout (Clear Syntax) and the project architecture (Clear Structure).

• Karl Broman’s website has a lots of good tutorial on repdoducible research. Of direct relevance to any researcher Organizing data into spreadsheet

• STAT545: The material of a brilliant course focusing on: “data wrangling, exploration, and analysis with R”.

• Special Cheers to the team behind the tidyverse-oriented course: Data Science for Ecology that seems pretty awesome.