top_n(10, -n) vs. btm_n(10, n)

[npjc]

I use count(group) %>% top_n(10, -n) type-of-thing lots... Future me would prefer if I could use count(group) %>% btm_n(10, n) instead as it is more easily understood at a glance.

From ?top_n():

# Find 10 players with most games
require("Lahman")
tbl_df(Batting) %>%
  group_by(playerID) %>%
  tally(G) %>%
  top_n(10)

I wish for:

# to find the 10 players with least games
# currently
... %>% top_n(10, -n)
# wishful behaviour
... %>% btm_n(10, n)

Or perhaps it is easier to just document examples of how you'd get the bottom n of something under top_n()

Agreed. I was just searching for how to do this and believe the alternative suggested (btm_n) would be preferred

[hadley]

But top_n() with a negative value doesn't work?

[hadley]

Ooooh, you're negating the variable n, not the argument n.

[hadley]

I think using -n to indicate that you want the bottom n values is a reasonable compromise between concision and expression.

[npjc]

I'm surprised. A quick survey of students learning R 2/15 were able to "read off" what was going to happen. But thanks for considering.