CODEWITHSUNDEEP
r
ABC
ABC

Reputation: 341

Ranking within each group

Consider the data:

set.seed(123)

x <- rbinom(12, 1, .5)  
y <- (x==0) * rexp(12, 1/100)
z <- (x==1) * rexp(12, 1/220)
group <- sample( rep(1:2, each=6) )

d <- data.frame(x, y, z, group)

Sorting the data first by y, then by z

d <- d[order(d$y,d$z),]

Now within each group, I want to give rank . The following codes work correctly:

ds <- split(d, d$group)

ds1 <- ds[[1]]
ds1$rank <- 1:nrow(ds1)

ds2 <- ds[[2]]
ds2$rank <- 1:nrow(ds2)

But without splitting the data frame, I want to rank within each group. How can I do that?

Upvotes: 4

Views: 3010

Answers (2)

akrun
akrun

Reputation: 887038

Here is an option using base R. We first order the dataset based on 'group', 'y', 'z' columns, then use ave to create the sequence by 'group'

d1 <- d[do.call(order, d[c("group", "y", "z")]),]
row.names(d1) <- NULL
d1$rank <- with(d1, ave(seq_along(group), group, FUN = seq_along))
d1
#   x         y          z group rank
#1  1   0.00000   6.988904     1    1
#2  1   0.00000 329.283431     1    2
#3  1   0.00000 353.287515     1    3
#4  0  35.51413   0.000000     1    4
#5  0  47.87604   0.000000     1    5
#6  0 272.62365   0.000000     1    6
#7  1   0.00000 212.491666     2    1
#8  1   0.00000 257.076377     2    2
#9  1   0.00000 326.760675     2    3
#10 1   0.00000 889.022577     2    4
#11 0  48.02147   0.000000     2    5
#12 0  84.97861   0.000000     2    6

Upvotes: 2

pe-perry
pe-perry

Reputation: 2621

dplyr way:

library(dplyr)
d %>% 
    arrange(group, y, z) %>% 
    group_by(group) %>% 
    mutate(rank = 1:n()) %>% 
    ungroup()

We first sort the data.frame by group then y and then z, then group it by group and then assign the rank for each observation.

Result:

# A tibble: 12 × 5
       x         y          z group  rank
   <int>     <dbl>      <dbl> <int> <int>
1      1   0.00000   6.988904     1     1
2      1   0.00000 329.283431     1     2
3      1   0.00000 353.287515     1     3
4      0  35.51413   0.000000     1     4
5      0  47.87604   0.000000     1     5
6      0 272.62365   0.000000     1     6
7      1   0.00000 212.491666     2     1
8      1   0.00000 257.076377     2     2
9      1   0.00000 326.760675     2     3
10     1   0.00000 889.022577     2     4
11     0  48.02147   0.000000     2     5
12     0  84.97861   0.000000     2     6

Upvotes: 4

Related Questions