vecvec

The vecvec package allows you to create vectors of mixed classes. Mixed type vectors are useful for data analysis packages as they allow semantically similar classes (such as time or distributions) to be stored within the same vector and operated upon efficiently with the same functions. It works similarly to factor(), in which the values of a vector (analogous to levels of a factor) are stored in the attributes. This also allows for efficient computation when the values are replicated, since operations apply to the unique values of the vector.

This package is intended for developers creating mixed type vector packages, and is unlikely useful for general data analysis (although I’d be curious if you find direct application of this tool, let me know how you use it).

Installation

You can install the stable version from CRAN:

You can install the development version of vecvec like so:

remotes::install_github("mitchelloharawild/vecvec")

Example

Use vecvec() to create a vecvec. Here we combine three vectors of different types.

library(vecvec)

x <- vecvec(as.Date("2025-03-29"), as.POSIXct("2025-01-03 10:05:00"), rnorm(10))
x
#> <vecvec[12]>
#>  [1] 2025-03-29          2025-01-03 10:05:00  1.262954285       
#>  [4] -0.326233361         1.329799263         1.272429321       
#>  [7]  0.414641434        -1.539950042        -0.928567035       
#> [10] -0.294720447        -0.005767173         2.404653389

Internally, the vectors are stored separately in the attributes.

str(x)
#>  vecvec [1:12] 2025-03-29, 2025-01-03 10:05:00,  1.262954285, -0.326233361,...
#> @ v:List of 3
#>  ..$ : Date[1:1], format: "2025-03-29"
#>  ..$ : POSIXct[1:1], format: "2025-01-03 10:05:00"
#>  ..$ : num [1:10] 1.263 -0.326 1.33 1.272 0.415 ...

Operations apply to all vector types.

x + 1
#> <vecvec[12]>
#>  [1] 2025-03-30          2025-01-03 10:05:01  2.26295428        
#>  [4]  0.67376664          2.32979926          2.27242932        
#>  [7]  1.41464143         -0.53995004          0.07143297        
#> [10]  0.70527955          0.99423283          3.40465339

When subsetting, unused values are dropped.

x[1]
#> <vecvec[1]>
#> [1] 2025-03-29
str(x[1])
#>  vecvec [1:1] 2025-03-29
#> @ v:List of 1
#>  ..$ : Date[1:1], format: "2025-03-29"

When a vecvec is replicated, only the indexing is replicated.

str(rep(x, each = 10))
#>  vecvec [1:120] 2025-03-29, 2025-03-29, 2025-03-29, 2025-03-29, 2025-03-29,...
#> @ v:List of 3
#>  ..$ : Date[1:1], format: "2025-03-29"
#>  ..$ : POSIXct[1:1], format: "2025-01-03 10:05:00"
#>  ..$ : num [1:10] 1.263 -0.326 1.33 1.272 0.415 ...
str(x[c(1,1,1,3,2,2)])
#>  vecvec [1:6] 2025-03-29, 2025-03-29, 2025-03-29,  1.262954285, 2025-01-03 ...
#> @ v:List of 3
#>  ..$ : Date[1:1], format: "2025-03-29"
#>  ..$ : num [1:10] 1.263 -0.326 1.33 1.272 0.415 ...
#>  ..$ : POSIXct[1:1], format: "2025-01-03 10:05:00"

The values in the attribute are unchanged, which allows for efficient computation since operations apply to the attributes (which are not replicated).

str(rep(x, each = 10) + 100)
#>  vecvec [1:120] 2025-07-07, 2025-07-07, 2025-07-07, 2025-07-07, 2025-07-07,...
#> @ v:List of 3
#>  ..$ : Date[1:1], format: "2025-07-07"
#>  ..$ : POSIXct[1:1], format: "2025-01-03 10:06:40"
#>  ..$ : num [1:10] 101.3 99.7 101.3 101.3 100.4 ...

If the arguments is differ across replicated indices, the resulting vecvec is expanded

str(rep(x, each = 10) + 1:120)
#>  vecvec [1:120] 2025-03-30, 2025-03-31, 2025-04-01, 2025-04-02, 2025-04-03,...
#> @ v:List of 3
#>  ..$ : Date[1:10], format: "2025-03-30" "2025-03-31" ...
#>  ..$ : POSIXct[1:10], format: "2025-01-03 10:05:11" "2025-01-03 10:05:12" ...
#>  ..$ : num [1:100] 22.3 23.3 24.3 25.3 26.3 ...