In dcomtois/summarytools: Tools to Quickly and Neatly Summarize Data

\definecolor{MidnightBlue}{HTML}{2E74B5}

library(knitr)
opts_chunk$set(comment = NA, 
               prompt  = FALSE,
               cache   = FALSE,
               echo    = TRUE,
               results = 'asis')
library(summarytools)

This document shows how to customize the content of the Stats / Values column in \color{MidnightBlue}{data frame summaries} generated using summarytools::dfSummary(). This feature was introduced in version 1.0.0 of summarytoolsas a response to a feature request that came up several times, in a \color{MidnightBlue}{form} or \color{MidnightBlue}{another}.

How it works

Two new options were created: dfSummary.custom.1 and dfSummary.custom.2. The first one has a predefined value -- it is the one that makes up the fourth row of the cell (showing IQR and CV). The second one is set to NA by default. If both options are defined (non-NA), the cell will now show 5 lines rather than 4, provided there are no additional line feed occurring within the cell, be it by design or by an "overflow" of one of the custom lines.

Setup & Baseline

We'll use the first column of iris to show results as they are before making any changes. But first, a little bit of setting-up:

\vspace{8pt}

library(summarytools)
st_options(plain.ascii     = FALSE,
           headings        = FALSE,
           footnote        = NA,
           round.digits    = 1,
           style           = "rmarkdown", # For freq(), descr(), & ctable()
           dfSummary.varnumbers   = FALSE,
           dfSummary.valid.col    = FALSE,
           dfSummary.silent       = TRUE,
           dfSummary.style        = "grid",
           tmp.img.dir            = "img")

Now let's show the default output: \vspace{8pt}

iris_subset <- iris[1]
dfSummary(iris_subset, graph.magnif = .45)

Example 1 : Removing IQR (CV)

Setting dfSummary.custom.1 to NA will remove the last line in Stats / Values: \vspace{8pt}

st_options(dfSummary.custom.1 = NA)
dfSummary(iris_subset, graph.magnif = .35) # Adjust graph size accordingly

Example 2 : Adding Q1 & Q3

Here we'll create the expression needed to generate new statistics, Q1 & Q3. The expression is evaluated while looping on column data, and we need to refer to that data. The variable name to use is, well, column_data. Another variable you can use is round.digits (we've set to 1 in the setup chunk on page 1). \vspace{8pt}

st_options(
  dfSummary.custom.1 = 
    expression(
      paste(
        "Q1 - Q3 :",
        round(
          quantile(column_data,
                   probs = .25,
                   type = 2, 
                   names = FALSE,
                   na.rm = TRUE),
          digits = round.digits
        ), " - ",
        round(
          quantile(column_data, 
                   probs = .75,
                   type = 2, 
                   names = FALSE,
                   na.rm = TRUE),
          digits = round.digits
        )
      )
    )           
)

dfSummary(iris_subset, graph.magnif = .45)

Example 3: Inserting Back IQR (CV)

It is always possible to reset the value of dfSummary.custom.1 to its initial value by using

st_options(dfSummary.custom.1 = "default")

But let's make things a bit more interesting by actually showing IQR (CV) under Q1 & Q3. For this, we will use the default expression for dfSummary.custom.1 to define dfSummary.custom.2:

\vspace{8pt}

st_options(
  dfSummary.custom.2 =
    expression(
      paste(
        paste0(
          trs("iqr"), " (", trs("cv"), ") : "
        ),
        format_number(
          IQR(column_data, na.rm = TRUE),
          round.digits
        ),
        " (",
        format_number(
          sd(column_data, na.rm = TRUE) /
              mean(column_data, na.rm = TRUE),
          round.digits
        ),
        ")",
        collapse = "", 
        sep = ""
    )
  )
)

dfSummary(iris[3:5], graph.magnif = .65) # Again, graph size adjusted

Don't forget to set na.rm = TRUE whenever necessary (most base R statistics use it with FALSE as default).

Number Formatting

You may have noticed that instead of round(), we used format_number(), which is a summarytools internal function. It applies not only rounding, but all relevant formatting attributes as well (nsmall, decimal.mark, big.mark, scientific,and so on).

Displaying Formatted Expressions

As shown in the \color{MidnightBlue}{Introduction to summarytools} vignette, the following bit of code can be used to retrieve and format the expressions stored in the custom options. To achieve good results, the chunk option results='markup' was used for this chunk.

st_options(dfSummary.custom.1 = "default")
formatR::tidy_source(
  text   = deparse(st_options("dfSummary.custom.1")),
  indent = 2,
  args.newline = TRUE
)