du

The goal of du is to implement the data usage metric.

Installation

You can install the released version of du from CRAN with:

And the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("muschellij2/du")

Example

This is a basic example which shows you how to solve a common problem:

if (requireNamespace("MASS", quietly = TRUE)) {
  utils::data(anorexia, package = "MASS")

  anorex.1 <- glm(Postwt ~ Prewt + Treat + offset(Prewt),
                  family = gaussian, data = anorexia[ anorexia$Prewt > 70.25,])
  sub_df = anorexia[ anorexia$Postwt < 80 , c("Prewt", "Postwt"), drop = FALSE]
  sub_df2 = anorexia[ anorexia$Prewt > 70.25,]
  print(head(disk_usage(anorex.1, full_data = anorexia)))
  print(head(du(anorex.1, sub_df, sub_df2,
                full_data = anorexia)))
}
#>      Treat Prewt Postwt
#> [1,]     1     1      1
#> [2,]     1     1      1
#> [3,]     1     1      1
#> [4,]     1     1      1
#> [5,]     1     1      1
#> [6,]     1     1      1
#>       Treat     Prewt    Postwt
#> 1 0.6666667 0.6666667 0.6666667
#> 2 0.6666667 0.6666667 0.6666667
#> 3 0.6666667 0.6666667 0.6666667
#> 4 0.6666667 0.6666667 0.6666667
#> 5 0.6666667 1.0000000 1.0000000
#> 6 0.6666667 1.0000000 1.0000000

Known Issues

The way things are found to be “used” is it is joined with the full data set. If the new subsetted object (e.g. data.frame) is not unique, then the join will fail as it does some extra checking.