R spells for data wizards

I've never come up with a good way for learning/teaching the cool parts of R. I feel like that's sort of how R is; there's an awesome way to do everything, but it's all very specific and kind of hacky.

I thought of some magic incantations that you might not find in introductory R books/documentation/classes and wrote about them below.


When loading a CSV, don't convert strings to factors.

read.csv('csvsoundsystem.com/soundsystem.csv', stringsAsFactors = F)

When writing a CSV, don't add the rownames.

write.csv(iris, file = 'iris.csv', row.names = F)


It's easy to miss a level of indexing, especially with lists.

str(list(a = 3)[1][[1]])
# num 3

str(list(a = 3)[1])
# List of 1
# $ a: num 3

str(list(a = 3))
# List of 1
# $ a: num 3

You can use character vectors indices.

# [1] "Black" "Brown" "Red"   "Blond"

row.names(HairEyeColor) <- c('Pink', 'Blue', 'Green', 'Clear')
#        Sex
# Eye     Male Female
#   Brown   32     36
#   Blue    11      9
#   Hazel   10      5
#   Green    3      2

#        Eye
# Hair    Brown Blue Hazel Green
#   Pink     32   11    10     3
#   Blue     53   50    25    15
#   Green    10   10     7     7
#   Clear     3   30     5     8


Factor levels are sorted alphabetically by default

# [1] "1"  "2"  "3"  "4"  "5"  "6"  "7"  "8"  "9"  "10"

If you want to change that, just create a new factor, specifying the level order manually.

factor(parking$GarOrLot, levels = c('G', 'L'))

And you rename a level or levels like so.

levels(OrchardSprays$treatment)[3:5] <- c('X', 'Y', 'Z')

Concatenating text

This is how you concatenate text.

paste('abc', 'def', sep = '')

In JavaScript, this would be 'abc' + 'def'. Sort of. R's paste is more powerful because supports vectors! If you pass it vectors, paste will ordinarily concatenate corresponding elements across vector.

paste(c('a','b','c'), 1:3)

If you want to concatenate the elements within a vector, use collapse.

paste(c('Pack', 'my', 'box', 'with', 'five', 'dozen', 'liquor', 'jugs.'), collapse = ' ')

In case that isn't clear, it would look like this in JavaScript:

['Pack', 'my', 'box', 'with', 'five', 'dozen', 'liquor', 'jugs.'].join(' ')


Show all factor levels in a ggplot.

ggplot(iris[1:50,]) + aes(x = Species, y = Sepal.Length) +
  scale_x_discrete('Species', drop = F) + geom_jitter()

Also, in general, use ggplot. Base R graphics are more work than they're worth, except maybe if you're making music videos.

That said, if you do use base R graphics, try using locator when you're perfecting the layout of base R graphics.


Update your packages.



Set your preferred mirror

r <- getOption("repos")
r["CRAN"] <- "http://cran.mirrors.hoobly.com"
options(repos = r)

Remove the carrots from the beginnings of the lines so that you can run code that you've copied from the shell.

options(prompt="  ")
options(continue="+ ")

Make the screen wide


Save your command history and output

sink(file = paste('~/.history/r-log-', strftime(Sys.time(), '%F %H:%M:%OS9'), '-', sep = ''), split=T)

Higher-order functions

R's "apply" functions would be called "maps" in other languages. If you're applying along a list or vector, lapply or sapply, respectively, are convenient.

apply maps along any dimension of an array; you specify the dimension as an argument.

mapply maps along multiple vectors, passing multiple arguments to the function.

rollapply is really cool. It applies a function with a rolling window. For example, here's a rolling z-score that Brian wrote.


roll_z <- function(x){
    scores <- z(x)

z_change <- rollapply(rnorm(1000), 40, roll_z)

Other stuff

Write packages. This is probably an okay guide.

ProjectTemplate is kind of okay too.


Use str to find out something's type. It stands for "structure".


sqldf works both on R data.frames and on other databases

sqldf('SELECT foo FROM bar') # Use the bar data.frame
sqldf('SELECT foo FROM bar', dbname = 'baz.db') # Use the baz.db SQLite database

Use download.file to download files.

Sort one thing by another thing.


More like this

R for people who already know something about programming
The introductory R texts and courses that I see normally read a bit differently from introductions to other languages. Introduction to [some language] for people who already know programming Som...
paul rand pizza box
After I made a modernist pizza box, I had 49 blank pizza boxes left-over. So I made this. And here's a vector mockup. After I made a modernist pizza box, I had 49 blank pizza boxes left-ov...
R graphics tricks that you probably shouldn't use
pre { font-size: 0.8em; } I've spent/wasted years of my life trying to make nice plots with base R graphics, so I've learned a bunch of base R graphics tricks. One annoying part of base R...