Group_by for consecutive patterns in R

Question

I'm trying to grouping based on consecutive patterns. This is the dataset.

num col1
1   SENSOR_01
2   SENSOR_05
3   SENSOR_05, SENSOR_07
4   SENSOR_05, SENSOR_07
5   SENSOR_07
6   SENSOR_05
7   SENSOR_01, SENSOR_03
8   SENSOR_01
9   SENSOR_03
10  SENSOR_01
11  SENSOR_05

structure(list(num = 1:11, col1 = structure(c(1L, 4L, 5L, 5L, 6L, 4L, 2L, 1L, 3L, 1L, 4L), .Label = c("SENSOR_01", "SENSOR_01, SENSOR_03", "SENSOR_03", "SENSOR_05", "SENSOR_05, SENSOR_07", "SENSOR_07" ), class = "factor")), class = "data.frame", row.names = c(NA, -11L))

If the row repeatably includes SENSOR_05 and SENSOR_07, it should be grouped. SENSOR_01 and SENSOR_03 set is similar. Here is my expected table (group field).

num col1                   group
1   SENSOR_01               1
2   SENSOR_05               2
3   SENSOR_05, SENSOR_07    2
4   SENSOR_05, SENSOR_07    2
5   SENSOR_07               2
6   SENSOR_05               2
7   SENSOR_01, SENSOR_03    3
8   SENSOR_01               3
9   SENSOR_03               3
10  SENSOR_01               3
11  SENSOR_05               4

This is my code, but It doesn't work well.

g1 <- c("SENSOR_05", "SENSOR_07")
g2 <- c("SENSOR_01", "SENSOR_03")

test %>%
  group_by(group = cumsum(col1 %in% (rep(c(g1, g2)))))

A. Suliman A. Suliman · Accepted Answer · 2019-07-28T08:58:53

library(dplyr)
df %>% mutate(flag=case_when(grepl(paste(g1,collapse = '|'),col1)~1,
                             grepl(paste(g2,collapse = '|'),col1)~2,
                             TRUE~3), 
              group=data.table::rleid(flag))

     num                 col1 flag group
  1    1            SENSOR_01    2     1
  2    2            SENSOR_05    1     2
  3    3 SENSOR_05, SENSOR_07    1     2
  4    4 SENSOR_05, SENSOR_07    1     2
  5    5            SENSOR_07    1     2
  6    6            SENSOR_05    1     2
  7    7 SENSOR_01, SENSOR_03    2     3
  8    8            SENSOR_01    2     3
  9    9            SENSOR_03    2     3
  10  10            SENSOR_01    2     3
  11  11            SENSOR_05    1     4

PS: I used SENSOR_05 or SENSOR_07 not SENSOR_05 and SENSOR_07

Group_by for consecutive patterns in R

1 Answers