Statistical transformations: layer_stats
Mix.install([
{:giggly, path: Path.join(__DIR__, "..")},
{:kino, "~> 0.19"}
])About this notebook
ggplot2's reference page for
layer_stats has no examples, so
this notebook says only what the page describes and where Giggly stands.
A layer's stat computes on its data before anything is drawn: the stat decides what is shown, and
the geom how. ggplot2 takes a layer's stat from a stat_*() function, from a geom's default, as a
string given to a geom (geom_line(stat = "density")), as a ggproto object, or through layer().
Some geoms are paired with a stat of their own, as geom_histogram() is with stat_bin() and
geom_boxplot() with stat_boxplot(), and most use stat_identity(), which leaves the data as it
is. The variables a stat computes can be mapped with after_stat(). Of ggplot2's stats, Giggly
computes these: Giggly.Plot.geom_bar/3 counts the rows at each x, or sums their weight, as
geom_bar() and stat_count() do; Giggly.Plot.geom_count/3 and Giggly.Plot.stat_sum/3 count
the rows at each position, as stat_sum() does; Giggly.Plot.geom_histogram/3,
Giggly.Plot.geom_freqpoly/3 and Giggly.Plot.stat_bin/3 count them in bins, as stat_bin() does;
Giggly.Plot.geom_boxplot/3 summarises each group in five numbers, as stat_boxplot() does;
Giggly.Plot.geom_area/3 aligns its groups at the same x values, as stat_align() does;
Giggly.Plot.geom_smooth/3 and Giggly.Plot.stat_smooth/3 fit a curve to each group, as
stat_smooth() does; and Giggly.Plot.geom_density/3 and Giggly.Plot.stat_density/3 estimate
each group's density, as stat_density() does. Its points, columns, paths, lines, rectangles,
ribbons and text draw the data as it is, as ggplot2's do with stat_identity(). A mapping reads
what a stat computes, such as counting's prop or binning's density, with
{:after_stat, [:prop]}. Not in Giggly yet: every other stat, stat_boxplot() and stat_align()
as functions of their own, giving a layer its stat by name, and layer().
