describe_all: Describe your data cleanly and effectively
In m-clark/tidyext: Tidy Extensions for Data Processing

Description Usage Arguments Details Value See Also Examples

Describe data sets with multiple variable types effectively.

describe_all(
  data,
  digits = 2,
  include_NAcat = TRUE,
  max_levels = 10,
  include_numeric = FALSE,
  NAcat_include = NULL,
  sort_by_freq = TRUE,
  ...
)

describe_all_num(data, digits = 2, ...)

describe_all_cat(
  data,
  digits = 2,
  include_NAcat = TRUE,
  max_levels = 10,
  include_numeric = FALSE,
  sort_by_freq = TRUE,
  as_ordered = FALSE
)

`data`	The dataset, of class data.frame.
`digits`	See [base::round()]. Default is 2, which for categorical is applied to the proportion (i.e. before converting to percentage).
`include_NAcat`	Include NA values as categorical levels? Default is `TRUE`.
`max_levels`	The maximum number of levels you want to display for categorical variables. Default is 10.
`include_numeric`	For categorical summary, also include numeric variables with unique values fewer or equal to `max_levels`? Default is `FALSE`.
`NAcat_include`	Deprecated alias of `include_NAcat`.
`sort_by_freq`	Sort categorical levels by frequency? Default is `TRUE`.
`...`	Additional arguments passed to `num_summary`
`as_ordered`	Return the categorical results with the levels as ordered. See details and example.

This function comes out of my frustrations from various data set summaries either being inadequate for my needs, too 'busy' with output, or unable to deal well with mixed data types.

Numeric data is treated separately from categorical, and provides the same information as in num_summary.

Categorical variables are defined as those with class 'character', 'factor', 'logical', 'ordered', combined with include_numeric. They are are summarized with frequencies and percentages. For empty categorical variables (e.g. after a subset), a warning is thrown. Note that max_levels is used with top_n, and so will return additional values when there are ties.

The as_ordered argument is to get around the notorious alphabetical ordering of ggplot. It returns a data.frame where the 'data' column contains the frequency information of the categorical levels, while leaving the levels in order (e.g. decreasing if sort_by_freq was TRUE). This way you can directly plot the result in the manner you've actually requested. See the example.

The functions describe_all_num and describe_all_cat will provide only numeric or only categorical data summaries respectively. describeAll is a deprecated alias.

A list with two elements of summaries for numeric and other variables respectively. Or the contents of those elements if the type-specific functions are used.

summary num_by num_summary

library(tidyext); library(dplyr)
X = data.frame(f1 =gl(2, 1, 20, labels=c('A', 'B')),
               f2=gl(2, 2, 20, labels=c('X', 'Q')))
X = X %>% mutate(bin1 = rbinom(20, 1, p=.5),
                 logic1 = sample(c(TRUE, FALSE), 20, replace = TRUE),
                 num1 = rnorm(20),
                 num2 = rpois(20, 5),
                 char1 = sample(letters, 20, replace = TRUE))
describe_all(X)

describe_all(data.frame(x=factor(1:7)), digits=5)
describe_all(mtcars, digits=5, include_numeric=TRUE, max_levels=3)

library(ggplot2)
freqs = describe_all_cat(mtcars,
                         include_numeric=TRUE,
                         max_levels=3,
                         as_ordered = TRUE)
freqs %>%
  filter(Variable == 'cyl') %>%
  tidyr::unnest() %>%
  ggplot(aes(x=Group, y=`%`)) +
  geom_point(size = 10)