Rob Weyant bio photo

Rob Weyant

Data Scientist at Powerley

Twitter LinkedIn Instagram Github Last.fm

Basics of magrittr in R

plot of chunk unnamed-chunk-1

The magrittr package offers a new operator that can help improve readability of your code, and make it easier to update and modify data wrangling code. The %>% operator has been adopted into dplyr and many of Hadley Wickham’s packages are written to be pipe-friendly.

The Problem

R code can get hard to read

sapply(iris[iris$Sepal.Length < mean(iris$Sepal.Length),-5],FUN = mean)

A (Possible) Solution - the pipe %>%

  • Similar to Unix pipe |
  • Code can be written in the order of execution, left to right
  • %>% will “pipe” information from one statement to the next
    • x %>% f is equivalent to f(x)
    • x %>% f(y) is equivalent to f(x,y)
    • x %>% f %>% g %>% h is equivalent to h(g(f(x)))

{magittr} provides 4 special operators

  • %>% - pipe operators
  • %T>% - tee operator
  • %$% - exposition operator
  • %<>% - compound assignment pipe operator

What %>% is doing

The %>% is taking the output of the left-hand side and using that for the first argument of the right-hand side, or where it finds a .

Basic Example

df <- data.frame(x1=rnorm(100),x2=rnorm(100),x3=rnorm(100))

df %>% head(1)  # same as using head(df,1)
##          x1        x2         x3
## 1 0.9836479 0.4554726 -0.3232914
df %>% head(.,1)  # same as using head(df,1)
##          x1        x2         x3
## 1 0.9836479 0.4554726 -0.3232914

A slightly more complicated example

library(ggplot2)
mtcars %>%
  xtabs(~gear+carb,data=.) %>%
  as.data.frame %>%
  ggplot(.,aes(x=gear,y=carb,size=Freq)) +
  geom_point()

plot of chunk unnamed-chunk-5

An even more complicated example

# Generate some sample data.
df <-
  data.frame(
    Price    = 1:100 %>% sample(replace = TRUE),
    Quantity = 1:10  %>% sample(replace = TRUE),
    Type     =
      0:1 %>%
      sample(replace = TRUE) %>%
      factor(labels = c("Buy", "Sell"))
  )

Source

The combination of %>% with {dplyr}

  • filter()
  • group_by()
  • summarise(),summarize()
  • arrange()
  • mutate()
  • select()
sapply(iris[iris$Sepal.Length < mean(iris$Sepal.Length),-5],FUN = mean)
## Sepal.Length  Sepal.Width Petal.Length  Petal.Width 
##      5.19875      3.13375      2.46250      0.66375
iris %>%
  mutate(avg.length=mean(Sepal.Length)) %>%
  filter(Sepal.Length<avg.length) %>%
  select(-Species,-avg.length) %>%
  summarise_each(funs(mean))
##   Sepal.Length Sepal.Width Petal.Length Petal.Width
## 1      5.19875     3.13375       2.4625     0.66375

%$% The exposition operator

  • Similar to with() or attach()
  • Useful for functions that don’t take a data parameter
table(CO2$Treatment,CO2$Type)
##             
##              Quebec Mississippi
##   nonchilled     21          21
##   chilled        21          21
# with(CO2,table(Treatment,Type))
CO2 %$% table(Treatment,Type)
##             Type
## Treatment    Quebec Mississippi
##   nonchilled     21          21
##   chilled        21          21

%T>% The Tee Operator

  • Allows a “break” in the pipe.
  • Executes right-hand side of %T>%, but will continue to pipe through to next statement
iris %>%
filter(Species != 'virginica') %>%
select(Sepal.Width,Sepal.Length) %T>%
plot %>%  # Make scatterplot and keep going
colMeans

plot of chunk unnamed-chunk-10

##  Sepal.Width Sepal.Length 
##        3.099        5.471

%<>% The Compound Assignment Operator

  • Combines a pipe and an assignment operator
  • Think i++ or x+=z from the C family, Python, Ruby, etc.
df <- rexp(5,.5) %>% data.frame(col1=.)
df
##        col1
## 1 1.9493899
## 2 0.1607936
## 3 0.1463735
## 4 2.0450395
## 5 0.3237476
df %<>% arrange(col1)
df
##        col1
## 1 0.1463735
## 2 0.1607936
## 3 0.3237476
## 4 1.9493899
## 5 2.0450395

More Resources

Basics of magrittr in R was published on and last modified on .