Run lengths encoding for missing values (NA)

Compute the lengths and indices of runs of NA in a vector – or the reverse operation.

na_rle(x = double(), index_by = seq_along(x), interval = NULL)

list_of_na_rle(x = double(), index_by = seq_along(x),
  interval = NULL)

na_rle_inverse(x)

na_rle_lengths(x)

na_rle_starts(x)

na_rle_ends(x)

Arguments

x

A vector.
index_by

A vector of the same length as x.
interval

if NULL, determined by the greatest common denominator; otherwise a supplied "interval" class. See ?tsibble::tsibble for details.

Value

An object of class rle_na or list_of_rle_na. A named list of:

  • lengths: the lengths of NA runs

  • indices: the starting indices of runs

Mathematical operations

Many math operations can be applied to objects returned from na_rle() and list_of_na_rle(), regarding the lengths of runs.

Examples

library(dplyr, warn.conflicts = FALSE)
df <- tibble(year = 2000:2019, temp = sample(0:30, size = 20))
df[c(1, 6, 13:16, 19), "temp"] <- NA
df
#> # A tibble: 20 x 2
#>     year  temp
#>    <int> <int>
#>  1  2000    NA
#>  2  2001    22
#>  3  2002    11
#>  4  2003     4
#>  5  2004     5
#>  6  2005    NA
#>  7  2006    25
#>  8  2007     3
#>  9  2008    26
#> 10  2009     8
#> 11  2010    27
#> 12  2011    20
#> 13  2012    NA
#> 14  2013    NA
#> 15  2014    NA
#> 16  2015    NA
#> 17  2016     6
#> 18  2017    10
#> 19  2018    NA
#> 20  2019    19

na_rle(df$temp) # indexed by the default positions
#> <Run Length Encoding <NA>[4]>
#> $lengths: <int> 1 1 4 1 
#> $indices: <int>  1  6 13 19
(x <- na_rle(df$temp, index_by = df$year)) # indexed by a variable
#> <Run Length Encoding <NA>[4]>
#> $lengths: <int> 1 1 4 1 
#> $indices: <int> 2000 2005 2012 2018

# getters
na_rle_inverse(x)
#> [1] 2000 2005 2012 2013 2014 2015 2018
na_rle_lengths(x)
#> [1] 1 1 4 1
na_rle_starts(x)
#> [1] 2000 2005 2012 2018
na_rle_ends(x)
#> [1] 2000 2005 2015 2018

# subsetting
x[1:2]
#> <Run Length Encoding <NA>[2]>
#> $lengths: <int> 1 1 
#> $indices: <int> 2000 2005

# math operations
length(x) # the number of runs
#> [1] 4
sum(x) # the total number of `NA`
#> [1] 7
range(x) # min & max runs
#> [1] 1 4

# list_of_na_rle() is useful when working with tabular data
na_rle_df <- df %>%
  mutate(group = rep(letters[1:2], each = 10)) %>%
  group_by(group) %>%
  summarise(na_runs = list_of_na_rle(temp, year))
na_rle_df
#> # A tibble: 2 x 2
#>   group         na_runs
#>   <chr> <list<rle<NA>>>
#> 1 a                 [2]
#> 2 b                 [2]

na_rle_inverse(na_rle_df$na_runs)
#> [[1]]
#> [1] 2000 2005
#> 
#> [[2]]
#> [1] 2012 2013 2014 2015 2018
#> 
sum(na_rle_df$na_runs)
#> [1] 2 5
range(na_rle_df$na_runs)
#> <list_of<integer>[2]>
#> [[1]]
#> [1] 1 1
#> 
#> [[2]]
#> [1] 1 4
#>