@summarize

Summarizing a dataset involves aggregating multiple rows down to (usually) a single row of data. This can be performed across the entire dataset, or if the dataset is grouped, then for each row in the dataset. This is implemented similarly to R's tidyverse using @summarize(). Out of admiration for Hadley Wickham, and to be consistent with the R tidyverse, both @summarize() and @summarise() are supported.

Note that summarization is different from other verbs in the Tidier.jl in 2 respects:

1. No auto-vectorization is performed when using @summarize()
2. One layer of grouping is removed after each @summarize() function.

If you require further changes to grouping beyond the defaults, you can either @ungroup() or call @group_by() to regroup by a different set of variables.

using Tidier
using RDatasets

movies = dataset("ggplot2", "movies");

Using @summarize() to count the number of movies in the dataset.

@chain movies begin
    @summarize(n = nrow())
end

1×1 DataFrame

Row	n
	Int64
1	58788

Using @summarize() to calculate average budget of movies in the dataset.

The median budget in this dataset is 3 million, and the mean budget is 13 million! Making movies must be way more lucrative than making Julia packages.

@chain movies begin
  @mutate(Budget = Budget / 1_000_000)
  @summarize(median_budget = median(skipmissing(Budget)),
             mean_budget = mean(skipmissing(Budget)))
end

1×2 DataFrame

Row	median_budget	mean_budget
	Float64	Float64
1	3.0	13.4125

Combining @group_by() with @summarise()

How many movies came out in each of the last 5 years?

@chain movies begin
  @group_by(Year)
  @summarise(n = nrow())
  @arrange(desc(Year))
  @slice(1:5)
end

5×2 DataFrame

Row	Year	n
	Int32	Int64
1	2005	349
2	2004	1945
3	2003	2158
4	2002	2168
5	2001	2121

Notice that there was no need to explicitly @ungroup() the dataset after summarizing here. The @summarise() function removed one layer of grouping. Since this dataset was only grouped by one variable (Year), it was no longer grouped after the @summarise was performed.

---

This page was generated using Literate.jl.