dplytab

A simple way to create multi-level statistic tables that integrates with the dplyr syntax.

Motivation

This is sorely missing from the tidyverse. There are a few multi-dimensional table creating packages and functions, notably ftable and the tables package. But I find this to be a more natural extension to the dplyr workflow process, and I need something that will easily knit to html.

General Description

So far its just a simple wrapper around some tidyr commands that reshape the data into the desired multi-table structure. Next step is to add/fix headers, and enhance the options....

Usage

For a brief example try the code below :

Based on iris - I add another 2 random dimensions:

irist <- cbind(iris
                ,dum = c(rep(0,75),rep(1,75))
                ,dum2 = c(rep("a",45),rep("b",105)))

Now lets use dplyr syntax to create a few statistics by the dimensions and pipe it along to dplyab:

irist %>%
  group_by(Species, dum, dum2) %>%
  summarise_at(vars(Sepal.Length), funs(n(),sd, mean)) %>%
  dplytab( row_by = c("dum2", "Species")
          ,col_by = c("dum"))

And this can easily extend to multiple analysis variables as well. For example:

irist %>%
  group_by(Species, dum, dum2) %>%
  summarise_at(vars(Sepal.Length, Sepal.Width), funs(n(),sd, mean)) %>%
  dplytab( row_by = c("dum2", "Species")
          ,col_by = c("dum"))

General Notes

• The table will be arranged in the order the variables are provided
• All Coloumn and row variables need to be provided (currently)

Improvements

Next stages include:

• Defining a class and print method for knitr
• Adding options to the order of the statistics displayed
• Adding totals and subtotals to row and/or coloumn statistics