table.glue

The goal of table.glue is to give you more control over the presentation of your data and also simplify the process of rounding and formatting data. The main idea is to create rounding specifications (starting with round_spec()) that can be plugged in, directly or through global options, to the table_glue() and table_value() functions.

Installation

You can install the development version from GitHub with:

# install.packages("remotes")
remotes::install_github("bcjaeger/table.glue")

Example

Suppose we want to write a summary statement about the mean and standard deviation (SD) of a column in the mtcars data. We’ll start by loading the glue, table.glue, and magrittr packages.

library(glue)       # for comparisons with table_glue
library(table.glue) # similar to glue but with built in rounding
library(magrittr)   # make rounding specifications using %>%

Use base R

The classic approach is to use base R functions format, round, and paste:

col_name <- "mpg"
col_mean <- mean(mtcars[[col_name]])
col_sd <- sd(mtcars[[col_name]])

col_mean_pretty <- format(signif(col_mean, digits = 3), nsmall = 1)
col_sd_pretty <- format(signif(col_sd, digits = 3), nsmall = 1)

paste0("The mean (SD) of ", col_name, " is ", 
       col_mean_pretty, " (", col_sd_pretty, ") ")
#> [1] "The mean (SD) of mpg is 20.1 (6.03) "

This gets the job done! Still, the code may be a little hard to read for a user who isn’t a grizzled paste() veteran. This is where the glue package really shines.

Use glue()

Instead of using paste(), glue() lets us write everything in one string, surrounding R object with curly brackets ( { } ) tells R that the glue() function should print the value of that R object rather than the raw string. For instance,

glue("the mean (SD) of {col_name} is {col_mean_pretty} ({col_sd_pretty})")
#> the mean (SD) of mpg is 20.1 (6.03)

This is certainly more readable and clean. The only thing glue() doesn’t do is make the pretty versions of col_mean and col_sd. This is where table.glue comes in.

Use table_glue()

The table.glue package lets you use glue() without having to make numbers pretty beforehand. For example, the code below uses table_glue(), one of the main functions in table.glue, to replicate the results we got from glue() but without using the pretty versions of col_mean and col_sd.

# notice that we are not using 'pretty' versions of col_mean and col_sd
table_glue("the mean (SD) of {col_name} is {col_mean} ({col_sd})")
#> [1] "the mean (SD) of mpg is 20 (6.0)"

Summary

• table_glue applies a general rounding convention to all numeric data that are supplied in the input string.

• The goal is to combine the clean syntax of glue() with a convenient and generally acceptable rounding specification.

Hopefully, most of your rounding needs will be met without going any further than this. However, the rabbit hole does go deeper. Let’s say you don’t like the default rounding specification and you want to make your own. You can do that!

rspec <- round_spec() %>% # make your own rounding specification
  round_using_signif(digits = 2) # round to 2 significant digits

# table glue adopts all the rules given by your specification
table_glue("the mean (SD) of {col_name} is {col_mean} ({col_sd})", 
           rspec = rspec)
#> [1] "the mean (SD) of mpg is 20 (6)"

More details on the different rounding specification rules are given in the package vignette: “Rounding specifications” (not written yet; in progress). One final nice feature of table.glue is the ability to make any rounding specification the default rounding specification used by all of the table_ functions, including table_glue:

# save your rounding specification, making it the default for table_glue
default_rounder_set(rspec)

# now you don't have to write rspec = rspec in table_glue()

table_glue("the mean (SD) of {col_name} is {col_mean} ({col_sd})")
#> [1] "the mean (SD) of mpg is 20 (6)"