dplyr join_by() argument

dplyr 1.1 introduced a new option for the by argument to the *_join() family of functions. You can now use join_by() instead of the previous c().

library(tidyverse)
library(gapminder)

dt_1 <-
  gapminder |>
  left_join(country_codes, by = join_by(country == country))

The short version without the named parameter works as well when using it as the third parameter in a *_join() function (the second parameter in a pipe.)

dt_2 <-
  gapminder |>
  left_join(country_codes, join_by(country == country))

It is an alternative to using c() in the argument.

dt_3 <-
  gapminder |>
  left_join(country_codes, by = c("country" = "country"))

The three created data frames are identical.

identical(dt_1, dt_2) & identical(dt_1, dt_3)

[1] TRUE

And here is a figure using the created data frame.

Code

set.seed(2)

pl_dt <-
  dt_1 |>
  filter(year == max(year)) |>
  select(country = iso_alpha, gdp_per_capita = gdpPercap, life_expectency = lifeExp)

ggplot(pl_dt, aes(x = gdp_per_capita, y = life_expectency)) +
  geom_point(color = "darkgrey") +
  scale_x_log10() +
  geom_smooth() +
  geom_text(
    data = sample_n(pl_dt, 10),
    aes(x = gdp_per_capita, y = life_expectency, label = country),
    vjust = -0.8, size = 3.5
  )

This is something I missed when reading the change log of the dplyr 1.1. release and did not find out about by reading the help pages of the *_join() functions.

It was different with the multiple matches revisions in dplyr 1.1. Here, I quickly run into the new warning message, explored and used it. To handle multiple matches more explicitly is really nice and I have started to use it regularly.

The recent post Teaching the tidyverse in 2023 provided a good review of the new features in the tidyverse released over the last months. I was aware of most of the changes and have started using the new approaches.

It is nice to see that the tidyverse is evolving and that the changes are documented so well.