Keep consecutive duplicates - r

Question

I have a data frame where one column contains some consecutive duplicates. I want to keep the rows with consecutive duplicates (any length >1). I would prefer a solution in dplyr or data.table.
Example data :
a <- seq(10,150,10)
b <- c("A", "A", "B", "C", "C", "A", "B", "B", "B", "C", "A", "C", "D", "E", "E")
df <- tibble(a, b)
Data:
# A tibble: 15 x 2
a b
<dbl> <chr>
1 10 A
2 20 A
3 30 B
4 40 C
5 50 C
6 60 A
7 70 B
8 80 B
9 90 B
10 100 C
11 110 A
12 120 C
13 130 D
14 140 E
15 150 E
So I would like to keep the rows with consecutive duplicates in column b.
Expected outcome:
# A tibble: 9 x 2
a b
<dbl> <chr>
1 10 A
2 20 A
4 40 C
5 50 C
7 70 B
8 80 B
9 90 B
14 140 E
15 150 E
Thanks!

Answer 1

Using the data.table input shown in the Note at the end, set N to be the number of elements in each group of consecutive elements and then keep groups for which it is greater than 1.
DT[, N :=.N, by = rleid(b)][N > 1, .(a, b)]
giving:
a b
1: 10 A
2: 20 A
3: 40 C
4: 50 C
5: 70 B
6: 80 B
7: 90 B
8: 140 E
9: 150 E
Note
We assume the input in reproducible form is:
library(data.table)
a <- seq(10,150,10)
b <- c("A", "A", "B", "C", "C", "A", "B", "B", "B", "C", "A", "C", "D", "E", "E")
DT <- data.table(a, b)

Answer 2

In dplyr we can use lag to create groups and select groups with more than 1 row.
library(dplyr)
df %>%
group_by(group = cumsum(b != lag(b, default = first(b)))) %>%
filter(n() > 1) %>%
ungroup() %>%
select(-group)
# a b
# <dbl> <chr>
#1 10 A
#2 20 A
#3 40 C
#4 50 C
#5 70 B
#6 80 B
#7 90 B
#8 140 E
#9 150 E
In base R, we can use rle and ave to subset rows from df
subset(df, ave(b, with(rle(b), rep(seq_along(values), lengths)), FUN = length) > 1)

Answer 3

Since you also have the data.table tag, i like using the data.table::rleid function for such tasks, i.e.
library(dplyr)
df %>%
group_by(grp = data.table::rleid(b), b) %>%
filter(n() > 1)
which gives,
# A tibble: 9 x 3
# Groups: grp, b [4]
a b grp
<dbl> <chr> <int>
1 10 A 1
2 20 A 1
3 40 C 3
4 50 C 3
5 70 B 5
6 80 B 5
7 90 B 5
8 140 E 10
9 150 E 10

Answer 4

You want to remove duplicate except when consecutive: the following code flags duplicate values and consecutive values, then keeps only rows that are not duplicate or that are part of a consecutive set of duplicates.
df %>%
mutate(duplicate = duplicated(b),
consecutive = c(NA, diff(as.integer(factor(b)))) == 0) %>%
filter(!duplicate | consecutive) %>%
select(-duplicate, -consecutive)

Answer 5

Use rle to get the run length.
Assuming df <- data.frame(a=a,b=b), then the following can make it
df[-cumsum(rle(b)$lengths)[rle(b)$lengths==1],]

Answer 6

Another solution utilizes both lead() and lag():
library(tidyverse)
a <- seq(10,150,10)
b <- c("A", "A", "B", "C", "C", "A", "B", "B", "B", "C", "A", "C", "D", "E", "E")
df <- tibble(a, b)
df %>% filter(b == lead(b) | b == lag(b))
#> # A tibble: 9 x 2
#> a b
#> <dbl> <chr>
#> 1 10 A
#> 2 20 A
#> 3 40 C
#> 4 50 C
#> 5 70 B
#> 6 80 B
#> 7 90 B
#> 8 140 E
#> 9 150 E
Created on 2019-10-21 by the reprex package (v0.3.0)

Answer 7

Here is another option (which should be faster):
D[-D[, {
x <- rowid(rleid(b)) < 2
.I[x & shift(x, -1L, fill=TRUE)]
}]]
timing code:
library(data.table)
set.seed(0L)
nr <- 1e7
nb <- 1e4
DT <- data.table(b=sample(nb, nr, TRUE))
#DT <- data.table(b=c("A", "A", "B", "C", "C", "A", "B", "B", "B", "C", "A", "C", "D", "E", "E"))
DT2 <- copy(DT)
mtd1 <- function(df) {
df[-cumsum(rle(b)$lengths)[rle(b)$lengths==1],]
}
mtd2 <- function(D) {
D[, N :=.N, by = rleid(b)][N > 1, .(b)]
}
mtd3 <- function(D) {
D[-D[, {
x <- rowid(rleid(b)) < 2
.I[x & shift(x, -1L, fill=TRUE)]
}]]
}
bench::mark(mtd1(DT), mtd2(DT2), mtd3(DT), check=FALSE)
timings:
# A tibble: 3 x 13
expression min median `itr/sec` mem_alloc `gc/sec` n_itr n_gc total_time result memory time gc
<bch:expr> <bch:tm> <bch:tm> <dbl> <bch:byt> <dbl> <int> <dbl> <bch:tm> <list> <list> <list> <list>
1 mtd1(DT) 1.1s 1.1s 0.908 1.98GB 10.9 1 12 1.1s <df[,1] [2,014 x ~ <df[,3] [59 x ~ <bch:t~ <tibble [1 x ~
2 mtd2(DT2) 2.88s 2.88s 0.348 267.12MB 0 1 0 2.88s <df[,1] [2,014 x ~ <df[,3] [23 x ~ <bch:t~ <tibble [1 x ~
3 mtd3(DT) 639.91ms 639.91ms 1.56 505.48MB 4.69 1 3 639.91ms <df[,1] [2,014 x ~ <df[,3] [24 x ~ <bch:t~ <tibble [1 x ~