Skip to content

Latest commit

 

History

History
25 lines (17 loc) · 949 Bytes

best_practices.md

File metadata and controls

25 lines (17 loc) · 949 Bytes
Raw

Best practices in R

Functions

  • Try to write pure functions
    • Avoid side-effects
    • Receive all arguments in the argument list (do not rely on global values)
    • Return only one data type
  • For complex functions, try separating them into pure and not pure
    • For example, dplyr processing is pure, but ggplot-ing is not pure
  • For Not pure functions (e.g., commands):
    • Have them return (invisibly) the original input, to facilitate pipelines

Packages

  • Always load them at the beginning of the (main) script
    • if the user does not have all the packages loaded, the script should fail fast
    • in a notebook or in a Rmd file, this is the "setup" chunk
    • if a package is used only with its namespace (e.g., tidyr::), list it in the setup chunk as a comment # library(tidyr)
  • Always use library() not require()