In jessezlotoff/rtabsplus: Tabulate One or Two Variables

#suppressPackageStartupMessages(library(rtabsplus))
library(devtools)
load_all()
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
suppressPackageStartupMessages(tab(iris,Species))

The rtabsplus package includes multiple versions of the tab function.

One-way Tab, Unweighted

The most basic usage is to tabulate proportions of a factor variable. Using the iris dataset as an example:

# tab function
#   usage: tab(df, v1) where df is a dataframe and v1 is the column name
tab(iris, Species)

This provides similar functionality as the built-in table function with a few key differences. It gives column proportions instead of case sizes and returns the results as a tibble. You can use the optional nsize parameter to include those values as well.

# tab function
tab(iris, Species, nsize=TRUE)

# compare to built-in table function
table(iris$Species)

There is also an option to include the upper and lower bounds of the 95% confidence interval. Although this is not survey data, it is calculated here for demonstration purposes.

# tab function
tab(iris, Species, nsize=TRUE, ci=TRUE)

The tab function also takes an optional parameter to create collapsed categories. Each new category is specified in a list as "new label"=c(inputs). Use "@auto" at the start of a label to auto-generate it.

# tab function
my_collapses = list("@auto" = c("versicolor","virginica"), "@auto2" = c("setosa"))
tab(iris, Species, collapses=my_collapses)

Two-way Tabs, Unweighted

The tab function can also handle two-way tabulations by specifying the optional v2 parameter.

# tab function
#   usage: tab(df, v1, v2=v2)
tab(iris, Species, v2=Petal.Width)

Again, the results are a tibble with column proportions, including Total. Additional columns such as "n_" (for nsize) or "lci_", "uci_" (for ci) use the standard prefixes before the specific column name.

Note, the order of the variables matters.

# Species vs Petal.Width
#   usage: tab(df, v1, v2=v2, nsize=FALSE)
tab(iris, Species, v2=Petal.Width)

# Petal.Width vs Species (will be different)
#   usage: tab(df, v1, v2=v2, nsize=FALSE)
tab(iris, Petal.Width, v2=Species)

Weighted Tabs

To calculate weighted tabs, use the optional weight_var or sdesign parameters. The first takes the name of a column with weights and the second takes a survey design object from the survey package.

Again, using the iris package even though it's not survey data:

tab(iris, Species, weight_var = Petal.Length)

Technical Notes

Full specification:

tab(df, v1, v2 = NULL, weight_var = NULL, sdesign = NULL, nsize = FALSE, ci = FALSE, to_factor=TRUE, collapses=list())

The function creates a temporary vv1, and if v2 is specified vv2, variable(s) in the data. The function checks that this/these columns are not present before running, otherwise printing an error with no output.

The stab function (think s-tab) works the same way except that it takes quoted inputs.

# s-tab version
stab(iris, "Species", v2="Petal.Width")

# equivalent tab version
tab(iris, Species, v2=Petal.Width)

The tab function wraps stab, which in turn wraps utab (for unweighted) and wtab (for weighted).

jessezlotoff/rtabsplus documentation built on May 26, 2019, 3:33 p.m.