Assign percentile for each element in a vector?

Asked Nov 16 '19 at 19:53

Active Nov 16 '19 at 19:53

Viewed 557 times

I have dataframe with a column "rt" that I would like to know what percentile each row falls under.

for example

set.seed ( 1)
df <- data.frame(rt = rnorm(60), id = rep(letters[1:3], rep(20)) )
> head ( df)
          rt id
1 -0.6264538  a
2  0.1836433  b
3 -0.8356286  c
4  1.5952808  a
5  0.3295078  b
6 -0.8204684  c

so the way I do it now is to loop through each quantile and assign the precentile for that number. For example

  q = quantile(df$rt, seq(0, 1, .01) )

    df[df$rt >= q[96], ]

        rt id
4  1.595281  a
11 1.511781  b
56 1.980400  b

I now know that these samples are in the top 96% and would just go through each row, however is there a way to do this for all numbers without manually going through each percentile?

thanks!

asked Nov 16 '19 at 19:53

Ahdee

4,679
4
34
58

1

Maybe something like https://stackoverflow.com/a/56723577/9524424 – Peter Nov 16 '19 at 20:01
Personally, I like the trick `df$percentile = order(order(df$rt)) / nrow(df)`. This gives you exact locations. – tmrlvi Nov 16 '19 at 20:06
What you want is the empirical cdf. `df$p <- ecdf(df$rt)(df$rt)` which looks a bit odd because it `ecdf()` returns a function (which makes plotting convenient). – Nov 16 '19 at 20:07
rank(df$rt)/nrow(df) ? – StupidWolf Nov 16 '19 at 20:17
1

@Peter thanks that did the trick. Here was the final code `df2 = df %>% mutate(PCT = ntile(rt, 100)) %>% data.frame()` – Ahdee Nov 16 '19 at 20:45
great, cheers!, – Peter Nov 16 '19 at 20:45

Assign percentile for each element in a vector?

0 Answers0