In halpo/tidymargins: Tidy Marginal Functions

library(printr)
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

tidymargins

The goal of tidymargins is to provide the with_margins() adverb function, as well as the related spread_each() function.

Installation

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

install.packages("tidymargins")

Basic Usage

with_margins is very simple to use. It is an adverb, i.e. it accepts and returns a function. The function, passed in, we'll call it original, should accept a table like data object as it's first argument and return a similar object. The resulting function, we'll call it modified, will also accept a data argument as it's first argument however that argument is expected to be a grouped data frame. Other arguments to modified are passed along to the original function. When modified is called with a grouped data frame original is called for every possible subset of grouping variables, including the full set of all original grouping variables and the empty set implying no grouping, The results are bound together into a single data frame and returned.
The example below shows this as an example.

library(dplyr)
library(tidymargins)
data(mtcars)

# our original function here is count
modified <- with_margins(count)

# modified is a function 
class(modified)

modified(group_by(mtcars, Cylinders = cyl, Gears = gear))

See that the output of modified is the counts grouped by; both cyl and gear, by cyl alone, by gear alone, and a grand total (no groups), all in a single data frame. When a variable is not used as a grouping variable it is replaced with the value "(All)", indicating that all levels of that variable are included. This label can be changes by specifying the all.name argument when calling with_margins, shown in the next example.

There is no need to store the intermediate function returned by with_margins, but since the returned function may also accept other arguments care should be taken to what is passing to which function. Care should also be taken with the pipe.

mtcars %>% 
    select(cyl, gear, mpg, disp, hp, qsec) %>% 
    group_by(cyl, gear) %>% 
    # <---------- with_margins ----------------><-- summarise --->
    with_margins(summarise_if, all.name='Total')(is.numeric, mean) %>% 
    # making pretty
    rename( "Miles/(US) gallon" = "mpg"
          , "Cylinders" = "cyl"
          , "Displacement (cu.in.)" = "disp"
          , "Gross horsepower" = "hp"
          , "1/4 mile time" = "qsec"
          , "Number of forward gears" = "gear"
          )