Key features:

• NA values are considered identical (unlike base::rle())
• Results returned as a data.frame (rather than a list), but still compatible with base::inverse.rle()
• Faster! Includes a C implementation for regular atomic vectors, and an R version compatible with every input base::rle() accepts.

What’s in the box

• rlenc() is C code for run-length encoding of raw, logical, integer, numeric and character vectors.
  • Groups NA values into a run (unlike base::rle())
  • Returns a data.frame rather than a list
  • Returned object is compatible with base::inverse.rle()
  • Can be 10x faster than base::rle()
• rlenc_compat()
  • A pure R version of rlenc() which is compatible with all inputs that base::rle() accepts
• rleid() returns an integer vector numbering the runs of identical values within a vector of numeric or character data. This is very similar to data.table::rleid(), execpt the data.table() version is much more configurable and flexible. This version is probably only useful if you wanted to avoid pulling in data.table as a dependency.

Installation

You can install from GitHub with:

# install.package('remotes')
remotes::install_github('coolbutuseless/purler')

ToDo

• Long vector support in rlenc()

rlenc() - run-length encoding output as a data.frame

input <- c(1, 1, 1, 2, 2, 8, 8, 8, 8, 8, NA, NA, NA, NA)

(result <- purler::rlenc(input))

  lengths values start
1       3      1     1
2       2      2     4
3       5      8     6
4       4     NA    11

inverse.rle(result)

 [1]  1  1  1  2  2  8  8  8  8  8 NA NA NA NA

rlenc() benchmark

library(tidyr)
library(bench)
library(dplyr)
library(ggplot2)

N <- 1000
M <- 10

zz <- sample(seq_len(M), N, replace = TRUE)

res <- bench::mark(
  rle(zz),
  rlenc(zz),
  rlenc_compat(zz),
  check = FALSE
)

plot(res) + theme_bw()

Run-length encoding with NAs

In base::rle(), runs of NA values are not treated as a group.

All functions in purler do treat NAs as identical for the purpose of creating groups

input <- c(1, 1, 2, NA, NA, NA, NA, 4, 4, 4)

base::rle(input)

Run Length Encoding
  lengths: int [1:7] 2 1 1 1 1 1 3
  values : num [1:7] 1 2 NA NA NA NA 4

purler::rlenc_compat(input)

  lengths values start
1       2      1     1
2       1      2     3
3       4     NA     4
4       3      4     8

purler::rlenc(input)

  lengths values start
1       2      1     1
2       1      2     3
3       4     NA     4
4       3      4     8

purler::rlenc_id(input)

 [1] 1 1 2 3 3 3 3 4 4 4

Run-length encoded group ids

rlenc_id() numbers the runs of identical values in a numeric or character vector.

For a more complete approach to this problem, see data.table::rleid()

input <- c(11, 11, 12, 12, 12, NA, NA, NA, NA)

rlenc_id(input)

[1] 1 1 2 2 2 3 3 3 3

Related Software

• base::rle()
• data.table::rleid()