dfSummary: Data frame Summary

View source: R/dfSummary.R

dfSummaryR Documentation

Data frame Summary


Summary of a data frame consisting of: variable names and types, labels if any, factor levels, frequencies and/or numerical summary statistics, barplots/histograms, and valid/missing observation counts and proportions.


  round.digits = 1,
  varnumbers = st_options("dfSummary.varnumbers"),
  labels.col = st_options("dfSummary.labels.col"),
  valid.col = st_options("dfSummary.valid.col"),
  na.col = st_options("dfSummary.na.col"),
  graph.col = st_options("dfSummary.graph.col"),
  graph.magnif = st_options("dfSummary.graph.magnif"),
  style = st_options("dfSummary.style"),
  plain.ascii = st_options("plain.ascii"),
  justify = "l",
  col.widths = NA,
  headings = st_options("headings"),
  display.labels = st_options("display.labels"),
  max.distinct.values = 10,
  trim.strings = FALSE,
  max.string.width = 25,
  split.cells = 40,
  split.tables = Inf,
  tmp.img.dir = st_options("tmp.img.dir"),
  keep.grp.vars = FALSE,
  silent = st_options("dfSummary.silent"),



A data frame.


Number of significant digits to display. Defaults to 1. Does not affect proportions, which always show 1 digit.


Logical. Show variable numbers in the first column. Defaults to TRUE. Can be set globally with st_options, option “dfSummary.varnumbers”.


Logical. If TRUE, variable labels (as defined with rapportools, Hmisc or summarytools' label functions, among others) will be displayed. TRUE by default, but the labels column is only shown if a label exists for at least one column. Can be set globally with st_options, option “dfSummary.labels.col”.


Logical. Include column indicating count and proportion of valid (non-missing) values. TRUE by default; can be set globally with st_options, option “dfSummary.valid.col”.


Logical. Include column indicating count and proportion of missing (NA) values. TRUE by default; can be set globally with st_options, option “dfSummary.na.col”.


Logical. Display barplots/histograms column. TRUE by default; can be set globally with st_options, option “dfSummary.graph.col”.


Numeric. Magnification factor for graphs column. Useful if the graphs show up too large (then use a value such as .75) or too small (use a value such as 1.25). Must be positive. Defaults to 1. Can be set globally with st_options, option “dfSummary.graph.magnif”.


Character. Argument used by pander. Defaults to “multiline”. The only other valid option is “grid”. Style “rmarkdown” will fallback to “multiline”.


Logical. pander argument; when TRUE, no markup characters will be used (useful when printing to console). Defaults to TRUE. Set to FALSE when in context of markdown rendering. To change the default value globally, see st_options.


String indicating alignment of columns; one of “l” (left) “c” (center), or “r” (right). Defaults to “l”.


Numeric or character. Vector of column widths. If numeric, values are assumed to be numbers of pixels. Otherwise, any CSS-supported units can be used. NA by default, meaning widths are calculated automatically.


Logical. Set to FALSE to omit headings. To change this default value globally, see st_options.


Logical. Should data frame label be displayed in the title section? Default is TRUE. To change this default value globally, see st_options.


The maximum number of values to display frequencies for. If variable has more distinct values than this number, the remaining frequencies will be reported as a whole, along with the number of additional distinct values. Defaults to 10.


Logical; for character variables, should leading and trailing white space be removed? Defaults to FALSE. See details section.


Limits the number of characters to display in the frequency tables. Defaults to 25.


A numeric argument passed to pander. It is the number of characters allowed on a line before splitting the cell. Defaults to 40.


pander argument which determines the maximum width of a table. Keeping the default value (Inf) is recommended.


Character. Directory used to store temporary images when rendering dfSummary() with 'method = "pander"', 'plain.ascii = TRUE' and 'style = "grid"'. See Details.


Logical. When using group_by, keep rows corresponding to grouping variable(s) in output table. When FALSE (default), variable numbers still reflect the the ordering in the full data frame (in other words, some numbers will be skipped in the variable number column).


Logical. Hide console messages. FALSE by default. To change this value globally, see st_options.


Additional arguments passed to pander.


The default value plain.ascii = TRUE is intended to facilitate interactive data exploration. When using the package for reporting with rmarkdown, make sure to set this option to FALSE.

When trim.strings is set to TRUE, trimming is done before calculating frequencies, be aware that those will be impacted accordingly.

Specifying tmp.img.dir allows producing results consistent with pandoc styling while also showing png graphs. Due to the fact that in Pandoc, column widths are determined by the length of cell contents even if said content is merely a link to an image, using standard R temporary directory to store the images would cause columns to be exceedingly wide. A shorter path is needed. On Mac OS and Linux, using “/tmp” is a sensible choice, since this directory is cleaned up automatically on a regular basis. On Windows however, there is no such convenient directory, so the user has to choose a directory and cleanup the temporary images manually after the document has been rendered. Providing a relative path such as “img”, omitting “./”, is recommended. The maximum length for this parameter is set to 5 characters. It can be set globally with st_options (e.g.: st_options(tmp.img.dir = ".").

It is possible to control which statistics are shown in the Stats / Values column. For this, see the Details and Examples sections of st_options.


A data frame with additional class summarytools containing as many rows as there are columns in x, with attributes to inform print method. Columns in the output data frame are:


Number indicating the order in which column appears in the data frame.


Name of the variable, along with its class(es).


Label of the variable (if applicable).

Stats / Values

For factors, a list of their values, limited by the max.distinct.values parameter. For character variables, the most common values (in descending frequency order), also limited by max.distinct.values. For numerical variables, common univariate statistics (mean, std. deviation, min, med, max, IQR and CV).

Freqs (% of Valid)

For factors and character variables, the frequencies and proportions of the values listed in the previous column. For numerical vectors, number of distinct values, or frequency of distinct values if their number is not greater than max.distinct.values.

Text Graph

An ASCII histogram for numerical variables, and ASCII barplot for factors and character variables.


An html encoded graph, either barplot or histogram.


Number and proportion of valid values.


Number and proportion of missing (NA and NAN) values.


Several packages provide functions for defining variable labels, summarytools being one of them. Some packages (Hmisc in particular) employ special classes for labelled objects, but summarytools doesn't use nor look for any such classes.


Dominic Comtois, dominic.comtois@gmail.com

See Also

label, print.summarytools


saved_x11_option <- st_options("use.x11")
st_options(use.x11 = FALSE)

# Exclude some of the columns to reduce table width
dfSummary(tobacco, varnumbers = FALSE, valid.col = FALSE)

# Limit number of categories to be displayed for categorical data
dfSummary(tobacco, max.distinct.values = 5, style = "grid")

# Using stby()
stby(tobacco, tobacco$gender, dfSummary)

st_options(use.x11 = saved_x11_option)

## Not run: 

# Show in Viewer or browser - no capital V in view(); stview() is also
# available in case of conflicts with other packages)

# Rmarkdown-ready
dfSummary(tobacco, style = "grid", plain.ascii = FALSE,
          varnumbers = FALSE, valid.col = FALSE, tmp.img.dir = "./img")

# Using group_by()
tobacco %>% group_by(gender) %>% dfSummary()

## End(Not run)

summarytools documentation built on May 20, 2022, 9:06 a.m.