My R cheatsheet

My R cheatsheet

Get the name of a variable within a function

deparse(substitute(x))

Useful for diagnostic messages ("x must be numeric") where you want the caller’s name for the argument, not the literal string "x".

Tidyverse equivalent: rlang::ensym(x) returns a symbol; rlang::as_label(rlang::enquo(x)) returns a string. Use these inside {rlang}-aware functions; stick with deparse(substitute()) for plain base R.

Pad numbers with leading zeroes

formatC(1:10, width = 3, flag = '0')
# "001" "002" ... "010"

Equivalent alternatives, in rough order of how often I reach for them:

sprintf("%03d", 1:10)              # base R, arguably more idiomatic
stringr::str_pad(1:10, 3, pad = '0')  # tidyverse
formatC(1:10, width = 3, flag = '0')  # what's documented above

Strip non-word characters from text

x <- '  abc <>?|!"£$%^&*()_+}{~@:¬}"  def  123'
gsub('\\W+', '', x)
# "abc_def123"

\W keeps underscores

\W matches non-word characters, where “word character” = [A-Za-z0-9_]. So underscores survive (visible in the output above). If you want strictly alphanumeric, use gsub('[^A-Za-z0-9]+', '', x) instead.

Set the reference level of a factor

# data.table syntax:
thedata[, newvar := relevel(factor(oldvar), 'Reference level here')]
 
# Equivalent base R:
thedata$newvar <- relevel(factor(thedata$oldvar), ref = 'Reference level here')
 
# forcats equivalent:
thedata$newvar <- forcats::fct_relevel(factor(thedata$oldvar), 'Reference level here')

Set contrasts globally

options(contrasts = rep('contr.treatment', 2))

The two elements of the vector apply to unordered and ordered factors respectively. R’s default is c("contr.treatment", "contr.poly") — i.e. polynomial contrasts for ordered factors, which can produce surprising regression coefficients if you weren’t expecting them. Setting both to treatment contrasts gives uniform “this level vs. reference” coefficients regardless of factor ordering.

Set this at the top of analysis scripts

Setting it inside a function won’t persist past the function’s exit unless you wrap it with on.exit(). For reproducibility, set globally near the script header, alongside set.seed().

---

LINKS