Different Pipe Operators

The dplyr Pipe: %>%

R’s native pipe operator, |>, was only introduced in R 4.1. Before that, we used the analogous operator provided by the dplyr package, %>%. The dplyr pipe is almost exactly the same as the base R version. For example, we can re-create our first motivating example as follows.

library(dplyr)

bfi[c("agree", "consc", "extra", "neuro", "open")] %>%
  cov(use = "pairwise") %>%
  diag() %>%
  sqrt()
    agree     consc     extra     neuro      open 
0.8984019 0.9513469 1.0609041 1.1963314 0.8083739

For most practical purposes, the Base R and dplyr pipes are interchangeable, but there are a few minor differences between |> and %>%. For example, the dplyr pipe uses the . character as the dataset placeholder symbol.

mtcars %>%
  lm(mpg ~ hp + disp, data = .) %>%
  summary()

Call:
lm(formula = mpg ~ hp + disp, data = .)

Residuals:
    Min      1Q  Median      3Q     Max 
-4.7945 -2.3036 -0.8246  1.8582  6.9363 

Coefficients:
             Estimate Std. Error t value Pr(>|t|)    
(Intercept) 30.735904   1.331566  23.083  < 2e-16 ***
hp          -0.024840   0.013385  -1.856 0.073679 .  
disp        -0.030346   0.007405  -4.098 0.000306 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 3.127 on 29 degrees of freedom
Multiple R-squared:  0.7482,    Adjusted R-squared:  0.7309 
F-statistic: 43.09 on 2 and 29 DF,  p-value: 2.062e-09

The dplyr pipe also allows more flexible usage of the dataset placeholder symbol.

mtcars %>%
  lm(mpg ~ hp + disp, data = ., weights = .$wt) %>%
  summary()

Call:
lm(formula = mpg ~ hp + disp, data = ., weights = .$wt)

Weighted Residuals:
   Min     1Q Median     3Q    Max 
-7.873 -3.313 -1.434  3.071 10.635 

Coefficients:
             Estimate Std. Error t value Pr(>|t|)    
(Intercept) 29.510142   1.329831  22.191  < 2e-16 ***
hp          -0.022962   0.011752  -1.954 0.060426 .  
disp        -0.027690   0.006297  -4.398 0.000135 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 5.114 on 29 degrees of freedom
Multiple R-squared:  0.7511,    Adjusted R-squared:  0.7339 
F-statistic: 43.75 on 2 and 29 DF,  p-value: 1.749e-09
mtcars |>
  lm(mpg ~ hp + disp, data = _, weights = _$wt) |>
  summary()
Error in summary(lm(mpg ~ hp + disp, data = mtcars, weights = "_"$wt)): invalid use of pipe placeholder (<input>:3:0)

The Exposition Pipe: %$%

The exposition pipe, %$%, is a very useful flavor or pipe operator provided by the magrittr package. The exposition pipe offers an elegant alternative to the dataset placeholder symbol in many common types of pipeline. Most notably, pipelines involving functions that employ the so-called formula interface, such as lm(), t.test(), or boxplot().

Instead of passing the processed dataset into the first argument of a downstream function, the exposition pipe exposes the column names of the processed data frame to the downstream function. For example, the following code uses the exposition pipe to regress age on extra in the bfi data.

library(magrittr)

bfi %$% lm(extra ~ age + open)

Call:
lm(formula = extra ~ age + open)

Coefficients:
(Intercept)          age         open  
   2.752007     0.004406     0.276073

The exposition pipe allows the lm() call to access the column names of bfi when resolving the model formula extra ~ age. We can achieve analogous results in several ways.

# Explicitly regress vectors:
lm(bfi$extra ~ bfi$age + bfi$open)

Call:
lm(formula = bfi$extra ~ bfi$age + bfi$open)

Coefficients:
(Intercept)      bfi$age     bfi$open  
   2.752007     0.004406     0.276073  
# Use the Base R equivalent of %$%:
with(bfi, lm(extra ~ age + open))

Call:
lm(formula = extra ~ age + open)

Coefficients:
(Intercept)          age         open  
   2.752007     0.004406     0.276073  
# Use a dataset placeholder symbol:
bfi |> lm(extra ~ age + open, data = _)

Call:
lm(formula = extra ~ age + open, data = bfi)

Coefficients:
(Intercept)          age         open  
   2.752007     0.004406     0.276073
Practice

Use the bfi dataset and an exposition pipe to create a pipeline to draw a conditional boxplot that visualizes the distribution of agree scores within gender groups.

  • Use the Base R boxplot() function to draw the plot.
library(magrittr)

bfi %$% boxplot(agree ~ gender)

Practice

Create a pipeline that includes at least on exposition pipe to perform the following operations with the bfi dataset:

  1. Randomly sample \(N = 500\) rows.
    • Use dplyr::slice_sample().
  2. Use the sampled rows to regress agree onto extra, open, and gender.
    • Use lm().
  3. Extract the residuals from the linear regression model.
  4. Compute the sum of the absolute values of the residuals.
library(dplyr)
set.seed(235711)

bfi |> 
  slice_sample(n = 500) %$%
  lm(agree ~ extra + open + gender) |>
  resid() |>
  abs() |>
  sum()
[1] 318.6229

The Assignment Pipe: %<>%

The final type of pipe that we’ll cover is certainly a niche tool, but it provides a nice little bit of syntactic sugar. We only care about assignment pipes in particular types of pipelines. Names, pipelines with two characteristics:

  1. The pipeline modifies a dataset
  2. The pipeline overwrites the original dataset with the modified version

For example, the following pipeline uses dplyr data manipulation functions to modify the bfi1 dataset and overwrite the original with the modified version produced by the pipeline.

bfi1 <- bfi
head(bfi1)
      a1 a2 a3 a4 a5 c1 c2 c3 c4 c5 e1 e2 e3 e4 e5 n1 n2 n3 n4 n5 o1 o2 o3 o4
61617  2  4  3  4  4  2  3  3  4  4  3  3  3  4  4  3  4  2  2  3  3  6  3  4
61618  2  4  5  2  5  5  4  4  3  4  1  1  6  4  3  3  3  3  5  5  4  2  4  3
61620  5  4  5  4  4  4  5  4  2  5  2  4  4  4  5  4  5  4  2  3  4  2  5  5
61621  4  4  6  5  5  4  4  3  5  5  5  3  4  4  4  2  5  2  4  1  3  3  4  3
61622  2  3  3  4  5  4  4  5  3  2  2  2  5  4  5  2  3  4  4  3  3  3  4  3
61623  6  6  5  6  5  6  6  6  1  3  2  1  6  5  6  3  5  2  2  3  4  3  5  6
      o5 gender    education age agree consc extra neuro open
61617  3   male         <NA>  16   4.0   2.8   3.8   2.8  3.0
61618  3 female         <NA>  18   4.2   4.0   5.0   3.8  4.0
61620  2 female         <NA>  17   3.8   4.0   4.2   3.6  4.8
61621  5 female         <NA>  17   4.6   3.0   3.6   2.8  3.2
61622  3   male         <NA>  17   4.0   4.4   4.8   3.2  3.6
61623  1 female some_college  21   4.6   5.6   5.6   3.0  5.0
bfi1 <- bfi1 %>%
  slice_sample(n = 20) %>%
  select(-matches("\\d$"))

head(bfi1)
  gender        education age agree consc extra neuro open
1   male  graduate_degree  38   5.2   5.8   5.2   1.0  4.8
2   male     some_college  27   4.2   5.4   3.8   3.0  3.4
3 female college_graduate  20   5.2   3.8   4.6   3.8  5.2
4   male     some_college  40   5.6   5.0   4.6   1.4  4.6
5   male     some_college  26   5.0   5.2   3.8   2.2  4.2
6 female college_graduate  27   5.6   5.0   4.8   1.6  5.0

The assignment pipe cleans up the syntax a little but and makes the destructive nature of the pipeline clear. When you see an assignment pipe, you know that the modified data produced by the pipeline will replace the original dataset.

bfi1 <- bfi
head(bfi1)
      a1 a2 a3 a4 a5 c1 c2 c3 c4 c5 e1 e2 e3 e4 e5 n1 n2 n3 n4 n5 o1 o2 o3 o4
61617  2  4  3  4  4  2  3  3  4  4  3  3  3  4  4  3  4  2  2  3  3  6  3  4
61618  2  4  5  2  5  5  4  4  3  4  1  1  6  4  3  3  3  3  5  5  4  2  4  3
61620  5  4  5  4  4  4  5  4  2  5  2  4  4  4  5  4  5  4  2  3  4  2  5  5
61621  4  4  6  5  5  4  4  3  5  5  5  3  4  4  4  2  5  2  4  1  3  3  4  3
61622  2  3  3  4  5  4  4  5  3  2  2  2  5  4  5  2  3  4  4  3  3  3  4  3
61623  6  6  5  6  5  6  6  6  1  3  2  1  6  5  6  3  5  2  2  3  4  3  5  6
      o5 gender    education age agree consc extra neuro open
61617  3   male         <NA>  16   4.0   2.8   3.8   2.8  3.0
61618  3 female         <NA>  18   4.2   4.0   5.0   3.8  4.0
61620  2 female         <NA>  17   3.8   4.0   4.2   3.6  4.8
61621  5 female         <NA>  17   4.6   3.0   3.6   2.8  3.2
61622  3   male         <NA>  17   4.0   4.4   4.8   3.2  3.6
61623  1 female some_college  21   4.6   5.6   5.6   3.0  5.0
bfi1 %<>%
  slice_sample(n = 20) %>%
  select(-matches("\\d$"))

head(bfi1)
  gender            education age agree consc extra neuro open
1 female         some_college  27   5.6   5.6   5.8   2.2  5.8
2 female         some_college  51   5.4   5.6   4.2   4.4  4.2
3   male high_school_graduate  49   4.0   3.4   4.0   3.0  4.4
4 female high_school_graduate  19   3.8   3.8   2.0   3.6  3.8
5 female         some_college  27   4.6   5.8   6.0   4.2  5.8
6 female         some_college  26   5.0   5.8   4.4   1.0  5.0

Note that we must use the dplyr pipe, %>%, if we include an assignment pipe in our pipeline. The Base R pipe doesn’t play nicely with the assignment pipe. More specifically, the assignment pipe behaves like a basic pipe (i.e., |> or %>%) when combined with the Base R pipe operator, as you can see below.

bfi1 <- bfi
head(bfi1)
      a1 a2 a3 a4 a5 c1 c2 c3 c4 c5 e1 e2 e3 e4 e5 n1 n2 n3 n4 n5 o1 o2 o3 o4
61617  2  4  3  4  4  2  3  3  4  4  3  3  3  4  4  3  4  2  2  3  3  6  3  4
61618  2  4  5  2  5  5  4  4  3  4  1  1  6  4  3  3  3  3  5  5  4  2  4  3
61620  5  4  5  4  4  4  5  4  2  5  2  4  4  4  5  4  5  4  2  3  4  2  5  5
61621  4  4  6  5  5  4  4  3  5  5  5  3  4  4  4  2  5  2  4  1  3  3  4  3
61622  2  3  3  4  5  4  4  5  3  2  2  2  5  4  5  2  3  4  4  3  3  3  4  3
61623  6  6  5  6  5  6  6  6  1  3  2  1  6  5  6  3  5  2  2  3  4  3  5  6
      o5 gender    education age agree consc extra neuro open
61617  3   male         <NA>  16   4.0   2.8   3.8   2.8  3.0
61618  3 female         <NA>  18   4.2   4.0   5.0   3.8  4.0
61620  2 female         <NA>  17   3.8   4.0   4.2   3.6  4.8
61621  5 female         <NA>  17   4.6   3.0   3.6   2.8  3.2
61622  3   male         <NA>  17   4.0   4.4   4.8   3.2  3.6
61623  1 female some_college  21   4.6   5.6   5.6   3.0  5.0
bfi1 %<>%
  slice_sample(n = 20) |>
  select(-matches("\\d$"))
   gender            education age agree consc extra neuro open
1    male     college_graduate  30   4.0   3.4   4.2   3.6  4.4
2  female high_school_graduate  41   5.4   5.0   5.2   3.6  3.6
3  female                 <NA>  16   4.6   4.6   2.6   3.6  4.0
4    male         some_college  18   3.0   3.2   4.8   1.6  5.4
5  female      graduate_degree  25   5.4   2.0   5.6   4.4  5.2
6  female         some_college  21   5.0   4.8   4.6   2.6  3.4
7  female     college_graduate  41   5.2   5.8   3.4   5.2  3.6
8  female         some_college  23   4.0   5.4   1.8   6.0  5.8
9  female     college_graduate  27   1.6   3.0   1.2   2.4  5.4
10 female         some_college  21   5.4   2.2   3.8   1.2  3.4
11 female         some_college  44   5.4   4.8   5.4   4.2  4.4
12 female         some_college  52   4.2   4.0   5.0   1.0  4.8
13 female         some_college  23   5.2   5.0   5.8   1.8  5.0
14 female         some_college  27   5.0   4.6   3.4   3.6  3.6
15   male         some_college  60   4.8   4.6   4.4   2.4  4.0
16 female high_school_graduate  25   5.0   4.4   5.2   5.0  4.2
17   male     college_graduate  26   4.4   4.0   5.4   1.4  4.4
18   male     some_high_school  35   5.0   4.6   4.0   3.2  5.4
19   male         some_college  24   4.0   3.2   4.6   2.0  4.2
20 female high_school_graduate  46   4.0   5.4   4.2   2.8  3.2
head(bfi1)
  a1 a2 a3 a4 a5 c1 c2 c3 c4 c5 e1 e2 e3 e4 e5 n1 n2 n3 n4 n5 o1 o2 o3 o4 o5
1  3  4  4  2  6  5  3  5  5  5  2  5  4  5  5  3  3  5  5  2  6  2  4  5  5
2  2  5  5  6  6  6  4  5  3  1  1  3  5  6  5  4  5  5  3  1  4  5  5  5  5
3  1  5  5  3  4  4  5  5  2  3  5  5  4  1  4  3  3  4  4  4  5  4  2  5  2
4  3  4  2  2  3  3  2  2  4  1  2  2  5  4  5  2  3  1  1  1  6  3  6  5  1
5  2  6  5  5  6  3  1  3  5  6  1  1  6  6  4  4  4  6  4  4  5  4  6  6  1
6  1  6  4  5  4  5  5  5  1  4  3  3  5  5  5  3  4  2  2  2  4  6  4  5  4
  gender            education age agree consc extra neuro open
1   male     college_graduate  30   4.0   3.4   4.2   3.6  4.4
2 female high_school_graduate  41   5.4   5.0   5.2   3.6  3.6
3 female                 <NA>  16   4.6   4.6   2.6   3.6  4.0
4   male         some_college  18   3.0   3.2   4.8   1.6  5.4
5 female      graduate_degree  25   5.4   2.0   5.6   4.4  5.2
6 female         some_college  21   5.0   4.8   4.6   2.6  3.4
Practice

Use the assignment pipe to create a pipeline that implements the following operations:

  1. Standardize every column in the mtcars dataset.
  2. Overwrites the original mtcars dataset with the standardized version.

You can use the scale() function to standardize variables.

library(magrittr)

mtcars %<>% scale()

We can sanity check the results by computing the means and variances of each variable.

colMeans(mtcars)
          mpg           cyl          disp            hp          drat 
 7.112366e-17 -1.474515e-17 -9.085614e-17  1.040834e-17 -2.918672e-16 
           wt          qsec            vs            am          gear 
 4.682398e-17  5.299580e-16  6.938894e-18  4.510281e-17 -3.469447e-18 
         carb 
 3.165870e-17 
apply(mtcars, 2, var)
 mpg  cyl disp   hp drat   wt qsec   vs   am gear carb 
   1    1    1    1    1    1    1    1    1    1    1

Looks good!

Back to top