Search code examples
rdataframerow

Select 3 rows which have the smallest value among all values in the column p

I have a data frame df. I would like to select 3 rows which have the smallest value in the column p.

df

     p      b
as   0.6    ab
yu   0.3    bc
hy   0.05   ak
get  0.7    ka

result

     p      b
as   0.6    ab
yu   0.3    bc
hy   0.05   ak
Solution

  • Two approaches:

    df[df$p <= sort(df$p)[3],]
#       p  b
# as 0.60 ab
# yu 0.30 bc
# hy 0.05 ak

    One problem with this is that when there are ties (for third) in p, you will get more than 3 rows. Also, this will not work well when there are fewer than 3 rows.

    Another approach, if you don't care about the order:

    head(df[order(df$p),], n = 3)

    which has the advantage that it will always give the minimum of 3 or the actual number of rows. One problem with this is that it will not tell you that there is a tie, it'll just cap the number of rows.

    (One could mitigate the re-ordering by adding a column with the pre-arranged order, then re-arrange on that column post head.)

    Over to you which flow makes more sense.

    Edit: an option that preserves order:

    df[ rank(df$p) < 4,]

    (inspired by @NotThatKindODr's suggested use of the ordered row_number() %in% 1:3)