How to label consecutive data points (clusters) as different groups

user_6648436 0
Viewed 65

I'd like to label the consecutive data points of same property (in this example "TRUE") as different groups (e.g., Group1, Group2, ...).

Below is the example data:

dt <- data.frame(value = c(TRUE, TRUE, FALSE, FALSE, TRUE, FALSE, TRUE, TRUE, TRUE, TRUE))

The "group" column is what I want to achive:

 value group
  TRUE    G1
  TRUE    G1
 FALSE  <NA>
 FALSE  <NA>
  TRUE    G2
 FALSE  <NA>
  TRUE    G3
  TRUE    G3
  TRUE    G3
  TRUE    G3
r
2 Answers

In base R, we can use rle

dt$group <- inverse.rle(within.list(rle(dt$value), {
          values <- NA^!values
     values[!is.na(values)] <- paste0("G", seq_along(values[!is.na(values)]))}))

-output

dt
   value group
1   TRUE    G1
2   TRUE    G1
3  FALSE  <NA>
4  FALSE  <NA>
5   TRUE    G2
6  FALSE  <NA>
7   TRUE    G3
8   TRUE    G3
9   TRUE    G3
10  TRUE    G3

or a bit more compact option

library(dplyr)
library(stringr)
dt %>% 
   mutate(group = str_c('G', cumsum(c(TRUE, diff(value) > 0)) * NA^!value))
   value group
1   TRUE    G1
2   TRUE    G1
3  FALSE  <NA>
4  FALSE  <NA>
5   TRUE    G2
6  FALSE  <NA>
7   TRUE    G3
8   TRUE    G3
9   TRUE    G3
10  TRUE    G3

or with rleid from data.table

library(data.table)
setDT(dt)[, tmp := rleid(value)][(value), 
     group := paste0("G", .GRP), tmp][, tmp := NULL][]
    value group
 1:  TRUE    G1
 2:  TRUE    G1
 3: FALSE  <NA>
 4: FALSE  <NA>
 5:  TRUE    G2
 6: FALSE  <NA>
 7:  TRUE    G3
 8:  TRUE    G3
 9:  TRUE    G3
10:  TRUE    G3
user_37322710

Another option with rle -

dt$group <-  paste0('G', with(rle(dt$value), rep(cumsum(values), lengths)))
dt$group[!dt$value] <- NA
dt

#   value group
#1   TRUE    G1
#2   TRUE    G1
#3  FALSE  <NA>
#4  FALSE  <NA>
#5   TRUE    G2
#6  FALSE  <NA>
#7   TRUE    G3
#8   TRUE    G3
#9   TRUE    G3
#10  TRUE    G3
user_39629140
Related