This vignette describes the dots+interval geoms and stats in
ggdist
. This is a flexible sub-family of stats and geoms
designed to make plotting dotplots straightforward. In particular, it
supports a selection of useful layouts (including the classic Wilkinson
layout, a weave layout, and a beeswarm layout) and can automatically
select the dot size so that the dotplot stays within the bounds of the
plot.
The following libraries are required to run this vignette:
geom_dotsinterval()
The dotsinterval
family of geoms and stats is a
sub-family of slabinterval (see vignette("slabinterval")
),
where the “slab” is a collection of dots forming a dotplot and the
interval is a summary point (e.g., mean, median, mode) with an arbitrary
number of intervals.
The base geom_dotsinterval()
uses a variety of custom
aesthetics to create the composite geometry:
Depending on whether you want a horizontal or vertical orientation,
you can provide ymin
and ymax
instead of
xmin
and xmax
. By default, some aesthetics
(e.g., fill
, color
, size
,
alpha
) set properties of multiple sub-geometries at once.
For example, the color
aesthetic by default sets both the
color of the point and the interval, but can also be overridden by
point_color
or interval_color
to set the color
of each sub-geometry separately.
Due to its relationship to the geom_slabinterval()
family, aesthetics specific to the “dots” sub-geometry are referred to
with the prefix slab_
. When using the standalone
geom_dots()
geometry, it is not necessary to use these
custom aesthetics:
geom_dotsinterval()
is often most useful when paired
with stat_dotsinterval()
, which will automatically
calculate points and intervals and map these onto endpoints of the
interval sub-geometry.
stat_dotsinterval()
and stat_dots()
can be
used on two types of data, depending on what aesthetic mappings you
provide:
Sample data; e.g. draws from a data
distribution, bootstrap distribution, Bayesian posterior distribution
(or any other distribution, really). To use the stats on sample data,
map sample values onto the x
or y
aesthetic.
Distribution objects and analytical
distributions. To use the stats on this type of data, you must
use the xdist
, or ydist
aesthetics, which take
distributional
objects, posterior::rvar()
objects, or distribution names
(e.g. "norm"
, which refers to the Normal distribution
provided by the dnorm/pnorm/qnorm
functions). When used on
analytical distributions
(e.g. distributional::dist_normal()
), the
quantiles
argument determines the number of quantiles used
(and therefore the number of dots shown); the default is
100
.
All dotsinterval
geoms can be plotted horizontally or
vertically. Depending on how aesthetics are mapped, they will attempt to
automatically determine the orientation; if this does not produce the
correct result, the orientation can be overridden by setting
orientation = "horizontal"
or
orientation = "vertical"
.
Size and layout of dots in the dotplot are controlled by four
parameters: scale
, binwidth
,
dotsize
, and stackratio
.
scale
: If binwidth
is not set (is
NA
), then the binwidth
is determined
automatically so that the height of the highest stack of dots is less
than scale
. The default value of scale
, 0.9,
ensures there is a small gap between dotplots when multiple dotplots are
drawn.
binwidth
: The width of the bins used to lay out the
dots:
NA
(default): Use scale
to determine bin
width.unit()
: the exact bin width to use.
If it is numeric
, the bin width is expressed in data units;
use unit()
to specify the width in terms of screen
coordinates (e.g. unit(0.1, "npc")
would make the bin width
0.1 normalized parent coordinates, which would be 10% of the
plot width.)unit()
s giving an acceptable
minimum and maximum width. The automatic bin width algorithm will
attempt to find the largest bin width between these two values that also
keeps the tallest stack of dots shorter than scale
.dotsize
: The size of the dots as a percentage of
binwidth
. The default value is 1.07
rather
than 1
. This value was chosen largely by trial and error,
to find a value that gives nice-looking layouts with circular dots on
continuous distributions, accounting for the fact that a slight overlap
of dots tends to give a nicer apparent visual distance between adjacent
stacks than the precise value of 1
.
stackratio
: The distance between the centers of dots
in a stack as a proportion of the height of each dot.
stackratio = 1
, the default, mean dots will just touch;
stackratio < 1
means dots will overlap each other, and
stackratio > 1
means dots will have gaps between
them.
The side
aesthetic allows you to adjust the positioning
and direction of the dots:
"top"
, "right"
, or
"topright"
: draw the dots on the top or on the right,
depending on orientation
"bottom"
, "left"
, or
"bottomleft"
: draw the dots on the bottom or on the left,
depending on orientation
"topleft"
: draw the dots on top or on the left,
depending on orientation
"bottomright"
: draw the dots on the bottom or on the
right, depending on orientation
"both"
: draw the dots mirrored, as in a “beeswarm”
plot.When orientation = "horizontal"
, this yields:
set.seed(1234)
x = rnorm(100)
side_plot = function(...) {
expand.grid(
x = x,
side = c("topright", "both", "bottomleft"),
stringsAsFactors = FALSE
) %>%
ggplot(aes(side = side, ...)) +
geom_dots() +
facet_grid(~ side, labeller = "label_both") +
labs(x = NULL, y = NULL) +
theme(panel.border = element_rect(color = "gray75", fill = NA))
}
side_plot(x = x) +
labs(title = "Horizontal geom_dots() with different values of side") +
scale_y_continuous(breaks = NULL)
When orientation = "vertical"
, this yields:
The layout
parameter allows you to adjust the algorithm
used to place dots:
"bin"
(default): places dots on the off-axis at the
midpoint of their bins as in the classic Wilkinson dotplot. This
maintains the alignment of rows and columns in the dotplot. This layout
is slightly different from the classic Wilkinson algorithm in that: (1)
it nudges bins slightly to avoid overlapping bins and (2) if the input
data are symmetrical it will return a symmetrical layout."weave"
: uses the same basic binning approach of “bin”,
but places dots in the off-axis at their actual positions (modulo
overlaps, which are nudged out of the way). This maintains the alignment
of rows but does not align dots within columns."hex"
: uses the same basic binning approach of “bin”,
but alternates placing dots +binwidth/4
or
-binwidth/4
in the off-axis from the bin center. This
allows hexagonal packing by setting a stackratio
less than
1
(something like 0.9
tends to work)."compactswarm"
layout from
beeswarm::beeswarm()
. Does not maintain alignment of rows
or columns, but can be more compact and neat looking, especially for
sample data (as opposed to quantile dotplots of theoretical
distributions, which may look better with "bin"
,
"weave"
, or "hex"
).When side
is "top"
, these layouts look like
this:
layout_plot = function(layout, side, ...) {
data.frame(
x = x
) %>%
ggplot(aes(x = x)) +
geom_dots(layout = layout, side = side, stackratio = if (layout == "hex") 0.9 else 1) +
labs(
subtitle = paste0("layout = ", deparse(layout), if (layout == "hex") " with stackratio = 0.9"),
x = NULL,
y = NULL
) +
scale_y_continuous(breaks = NULL) +
theme(panel.border = element_rect(color = "gray75", fill = NA))
}
(layout_plot("bin", side = "top") + layout_plot("hex", side = "top")) /
(layout_plot("weave", side = "top") + layout_plot("swarm", side = "top")) +
plot_annotation(title = 'geom_dots() layouts with side = "top"')
When side
is "both"
, these layouts look
like this:
(layout_plot("bin", side = "both") + layout_plot("hex", side = "both")) /
(layout_plot("weave", side = "both") + layout_plot("swarm", side = "both")) +
plot_annotation(title = 'geom_dots() layouts with side = "both"')
Thus, it is possible to create beeswarm plots by using
geom_dots()
with side = "both"
:
set.seed(1234)
abc_df = tibble(
value = rnorm(300, mean = c(1,2,3), sd = c(1,2,2)),
abc = rep(c("a", "b", "c"), 100)
)
abc_df %>%
ggplot(aes(x = abc, y = value)) +
geom_dots(side = "both") +
ggtitle('geom_dots(side = "both")')
side = "both"
also tends to work well with the
"hex"
and "swarm"
layouts for more
classic-looking “beeswarm” plots:
abc_df %>%
ggplot(aes(x = abc, y = value)) +
geom_dots(side = "both", layout = "hex", stackratio = 0.92) +
ggtitle('geom_dots(side = "both", layout = "hex")')
The combination of binwidth = unit(1.5, "mm")
and
overflow = "compress"
(see the section on large samples,
below) can be used to set the dot size to a specific size while
guaranteeing the layout stays within the bounds of the geom. This
combination is used by two shortcut geoms, geom_swarm()
and
geom_weave()
, which use the "swarm"
and
"weave"
layouts respectively. These also use
side = "both"
, and are intended to make it easy to create
good-looking beeswarm plots without manually tweaking
settings:
set.seed(1234)
swarm_data = tibble(
y = rnorm(300, c(1,4)),
g = rep(c("a","b"), 150)
)
swarm_plot = swarm_data %>%
ggplot(aes(x = g, y = y)) +
geom_swarm(linewidth = 0, alpha = 0.75) +
labs(title = "geom_swarm()")
weave_plot = swarm_data %>%
ggplot(aes(x = g, y = y)) +
geom_weave(linewidth = 0, alpha = 0.75) +
labs(title = "geom_weave()")
swarm_plot + weave_plot
color
, fill
, shape
,
and linewidth
Aesthetics like color
, fill
,
shape
, and linewidth
can be varied over the
dots. For example, we can vary the fill
aesthetic to create
two subgroups, and use position = "dodge"
to dodge entire
“swarms” at once so the subgroups do not overlap. We’ll also set
linewidth = 0
so that the default gray outline is not
drawn:
set.seed(12345)
abcc_df = tibble(
value = rnorm(300, mean = c(1,2,3,4), sd = c(1,2,2,1)),
abc = rep(c("a", "b", "c", "c"), 75),
hi = rep(c("h", "h", "h", "i"), 75)
)
abcc_df %>%
ggplot(aes(y = value, x = abc, fill = hi)) +
geom_weave(position = "dodge", linewidth = 0, alpha = 0.75) +
scale_fill_brewer(palette = "Dark2") +
ggtitle(
'geom_weave(position = "dodge")',
'aes(fill = hi, shape = hi)'
)
By default, if you assign a discrete variable to fill
,
color,
shape, etc it will also be used in the
group`
aesthetic to determine dot groups, which are laid out separate (and can
be dodged separately, as above).
If you override this behavior by setting group
to
NA
(or to some other variable you want to group dot layouts
by), geom_dotsinterval()
will leave dots in data order
within the layout but allow aesthetics to vary across them.
For example:
abcc_df %>%
ggplot(aes(y = value, x = abc, fill = hi, group = NA)) +
geom_dots(linewidth = 0) +
scale_color_brewer(palette = "Dark2") +
ggtitle(
'geom_dots()',
'aes(fill = hi, group = NA)'
)
By default, dot positions within bins for the "bin"
layout are determined by their data values (e.g. by the y
values in the above chart). You can override this by passing a variable
to the order
aesthetic, which will set the sort order
within bins. This can be used to create “stacked” dotplots by setting
order
to a discrete variable:
Continuous variables can also be varied within groups. Since
continuous variables will not automatically set the group
aesthetic, we can simply assign them to the desired aesthetic we want to
vary:
When sample sizes can vary widely (and dynamically), it can be difficult to set a reasonable dot size that works on all charts. In this case, it can be useful to set constraints on the dot sizes picked by the automatic bin width selection algorithm.
For example, on very large samples, dots may become smaller than desired. Consider the following increasingly large samples:
set.seed(1234)
ns = c(50, 200, 500, 5000)
increasing_samples = data.frame(
x = rgamma(sum(ns), 2, 2),
n = rep(ns, ns)
)
increasing_samples %>%
ggplot(aes(x = x)) +
geom_dots() +
facet_wrap(~ n) +
labs(
title = "geom_dots()",
subtitle = "on large samples, dots may get too small"
)
The dots become quite small on the 5000-dot dotplot, making it harder to read.
You can set constraints on the desired dot size / bin width by using
the binwidth
argument. To set a specific bin width, pass a
single value; to set constraints, pass a length-2 vector, where the
first element is the min and the second the max. The min can be
0
and the max can be Inf
if you only want to
constrain the other value (max or min, respectively). The bin width can
be in data units (using numeric
values) or in plotting
units (using grid::unit()
s).
For example, we could constrain the dot size to be greater than 1mm:
increasing_samples %>%
ggplot(aes(x = x)) +
geom_dots(binwidth = unit(c(1, Inf), "mm")) +
facet_wrap(~ n) +
labs(
title = "geom_dots()",
subtitle = 'binwidth = unit(c(1.5, Inf), "mm")'
)
## Warning: The provided binwidth will cause dots to overflow the boundaries of the geometry.
## → Set `binwidth = NA` to automatically determine a binwidth that ensures dots fit within the
## bounds,
## → OR set `overflow = "compress"` to automatically reduce the spacing between dots to ensure the
## dots fit within the bounds,
## → OR set `overflow = "keep"` to allow dots to overflow the bounds of the geometry without producing
## a warning.
## ℹ For more information, see the documentation of the `binwidth` and `overflow` arguments of
## `?ggdist::geom_dots()` or the section on constraining dot sizes in vignette("dotsinterval")
## (`vignette(ggdist::dotsinterval)`).
## The provided binwidth will cause dots to overflow the boundaries of the geometry.
## → Set `binwidth = NA` to automatically determine a binwidth that ensures dots fit within the
## bounds,
## → OR set `overflow = "compress"` to automatically reduce the spacing between dots to ensure the
## dots fit within the bounds,
## → OR set `overflow = "keep"` to allow dots to overflow the bounds of the geometry without producing
## a warning.
## ℹ For more information, see the documentation of the `binwidth` and `overflow` arguments of
## `?ggdist::geom_dots()` or the section on constraining dot sizes in vignette("dotsinterval")
## (`vignette(ggdist::dotsinterval)`).
## The provided binwidth will cause dots to overflow the boundaries of the geometry.
## → Set `binwidth = NA` to automatically determine a binwidth that ensures dots fit within the
## bounds,
## → OR set `overflow = "compress"` to automatically reduce the spacing between dots to ensure the
## dots fit within the bounds,
## → OR set `overflow = "keep"` to allow dots to overflow the bounds of the geometry without producing
## a warning.
## ℹ For more information, see the documentation of the `binwidth` and `overflow` arguments of
## `?ggdist::geom_dots()` or the section on constraining dot sizes in vignette("dotsinterval")
## (`vignette(ggdist::dotsinterval)`).
Notice how the dots now go off the page and we receive a warning with
suggestions on how to fix the layout. If we set
overflow = "compress"
, instead of overflowing, the layout
will compress the spacing between dots to keep them within the
geometry’s bounds:
increasing_samples %>%
ggplot(aes(x = x)) +
geom_dots(binwidth = unit(c(1, Inf), "mm"), overflow = "compress", alpha = 0.75) +
facet_wrap(~ n) +
labs(
title = "geom_dots()",
subtitle = 'binwidth = unit(c(1, Inf), "mm"), overflow = "compress"'
)
These settings give reasonable displays in small sample sizes and scale up to larger sample sizes without changing settings.
The dots family includes a variety of features to make visualizing discrete and categorical distributions easier. These distributions can be hard to visualize under the default settings if the dots become very small:
set.seed(1234)
abcd_df = tibble(
x = sample(c("a", "b", "c", "d"), 1000, replace = TRUE, prob = c(0.27, 0.6, 0.03, 0.005)),
g = rep(c("a","b"), 500)
)
abcd_df %>%
ggplot(aes(x = x)) +
geom_dots() +
scale_y_continuous(breaks = NULL) +
labs(
title = "geom_dots()",
subtitle = "on a large discrete sample"
)
The automatic bin width algorithm selects a dot size that is very small in order to ensure the tallest bin fits in the plot, but this means the dots are hard to see.
Bar-like layouts can be achieved by using
layout = "bar"
:
abcd_df %>%
ggplot(aes(x = x, fill = g, order = g)) +
geom_dots(layout = "bar", group = NA, color = NA) +
scale_y_continuous(breaks = NULL) +
labs(
title = 'geom_dots(aes(fill = g), layout = "bar", group = NA)',
subtitle = "on a large discrete sample"
)
Notice how we set group = NA
to override the default
ggplot2
behavior of grouping data by all discrete
variables. This allows the layout to be calculated taking all groups
into account.
We can also use the smooth
parameter to improve the
display of discrete distributions, for which geom_dots()
supports a handful of smoothers. These all correspond to
functions that start with smooth_
, like
smooth_bounded()
, smooth_unbounded()
, and
smooth_discrete()
, and can be applied either by passing the
suffix as a string (e.g. smooth = "bounded"
) or by passing
the function itself, to set specific options on it
(e.g. smooth = smooth_bounded(adjust = 0.5)
).
smooth_discrete()
applies a kernel density smoother
whose default bandwidth is less than the distances between bins. We can
use the kernel
argument (passed to
density_bounded()
; the same kernels from
stats::density()
are available) to change the shape of the
bins.
For example, using the "epanechnikov"
(parabolic) kernel
along with side = "both"
, we can create lozenge-like
shapes. We’ll abbreviate the kernel "ep"
to save typing out
"epanechnikov"
(partial matching is allowed):
Like the stat_slabinterval()
family,
stat_dotsinterval()
and stat_dots()
support
using both sample data (via x
and y
aesthetics) or analytical distributions (via the xdist
and
ydist
aesthetics). For analytical distributions, these
stats accept specifications for distributions in one of two ways:
Using distribution names as character vectors: this format uses aesthetics as follows:
xdist
, ydist
, or dist
: the
name of the distribution, following R’s naming scheme. This is a string
which should have "p"
, "q"
, and
"d"
functions defined for it: e.g., “norm” is a valid
distribution name because the pnorm()
,
qnorm()
, and dnorm()
functions define the CDF,
quantile function, and density function of the Normal distribution.args
or arg1
, … arg9
:
arguments for the distribution. If you use args
, it should
be a list column where each element is a list containing arguments for
the distribution functions; alternatively, you can pass the arguments
directly using arg1
, … arg9
.Using distribution vectors from the distributional
package or posterior::rvar()
objects: this format
uses aesthetics as follows:
xdist
, ydist
, or dist
: a
distribution vector or posterior::rvar()
produced by
functions such as distributional::dist_normal()
,
distributional::dist_beta()
,
posterior::rvar_rng()
, etc.For example, here are a variety of distributions:
dist_df = tibble(
dist = c(dist_normal(1,0.25), dist_beta(3,3), dist_gamma(5,5)),
dist_name = format(dist)
)
dist_df %>%
ggplot(aes(y = dist_name, xdist = dist)) +
stat_dotsinterval(subguide = 'integer') +
ggtitle(
"stat_dotsinterval(subguide = 'integer')",
"aes(y = dist_name, xdist = dist)"
)
This example also shows the use of sub-guides to label dot counts.
See the documentation of subguide_axis()
and its shortcuts
(particularly subguide_integer()
and
subguide_count()
) for more examples.
Analytical distributions are shown by default using 100 quantiles,
sometimes referred to as a quantile dotplot, which can help
people make better decisions under uncertainty (Kay 2016, Fernandes 2018). This
can be changed using the quantiles
argument. For example,
we can plot the same distributions again using 1000 quantiles. We’ll
also make use of point_interval
to plot the mode and
highest-density continuous intervals (instead of the default median and
quantile intervals; see point_interval()
).
We’ll also highlight some intervals by coloring the dots. Like with
the stat_slabinterval()
family, computed variables from the
interval sub-geometry (level
and .width
) are
available to the dots/slab sub-geometry, and correspond to the smallest
interval containing that dot. We can use these to color dots according
to the interval containing them (we’ll also use the "weave"
layout since it maintains x positions better than the "bin"
layout):
dist_df %>%
ggplot(aes(y = dist_name, xdist = dist, slab_fill = after_stat(level))) +
stat_dotsinterval(quantiles = 1000, point_interval = mode_hdci, layout = "weave", slab_color = NA) +
scale_color_manual(values = scales::brewer_pal()(3)[-1], aesthetics = "slab_fill") +
ggtitle(
"stat_dotsinterval(quantiles = 1000, point_interval = mode_hdci)",
"aes(y = dist_name, xdist = dist, slab_fill = after_stat(level))"
)
When summarizing sample distributions with
stat_dots()
/stat_dotsinterval()
(e.g. samples
from Bayesian posteriors), one can also use the quantiles
argument, though it is not on by default.
While varying discrete aesthetics works similarly with
stat_dotsinterval()
/stat_dots()
as it does
with geom_dotsinterval()
/geom_dots()
, varying
continuous aesthetics within dot groups typically requires mapping the
continuous aesthetic after the stats are computed. This is
because the stat (at least for analytical distributions) must first
generate the quantiles before properties of those quantiles can be
mapped to aesthetics.
Thus, because it relies upon generated variables from the stat, you
can use the after_stat()
or stage()
functions
from ggplot2
to map those variables. For example:
dist_df %>%
ggplot(aes(y = dist_name, xdist = dist, slab_color = after_stat(x))) +
stat_dotsinterval(slab_shape = 19, quantiles = 500) +
scale_color_distiller(aesthetics = "slab_color", guide = "colorbar2") +
ggtitle(
"stat_dotsinterval(slab_shape = 19, quantiles = 500)",
'aes(slab_color = after_stat(x)) +\nscale_color_distiller(aesthetics = "slab_color", guide = "colorbar2")'
)
This example also demonstrates the use of sub-geometry scales: the
slab_
-prefixed aesthetics slab_color
and
slab_shape
must be used to target the color and shape of
the slab (“slab” here refers to the stack of dots) when using
geom_dotsinterval()
and stat_dotsinterval()
to
disambiguate between the point/interval and the dot stack. When using
stat_dots()
/geom_dots()
this is not
necessary.
Also note the use of scale_color_distiller()
, a base
ggplot2 color scale, with the slab_color
aesthetic by
setting the aesthetics
and guide
properties
(the latter is necessary because the default
guide = "colorbar"
will not work with non-standard color
aesthetics).
Another potentially useful application of post-stat aesthetic
computation is to apply thresholds on a dotplot, coloring points on one
side of a line differently. However, the default dotplot layout,
"bin"
, can cause dots to be on the wrong side of a cutoff
when coloring dots within dotplots. Thus it can be useful when plotting
thresholds to use the "weave"
or "swarm"
layouts, which tend to position dots closer to their true x
positions, rather than at bin centers:
ab_df = tibble(
ab = c("a", "b"),
mean = c(5, 7),
sd = c(1, 1.5)
)
ab_df %>%
ggplot(aes(y = ab, xdist = dist_normal(mean, sd), fill = after_stat(x < 6))) +
stat_dots(position = "dodge", color = NA, layout = "weave") +
labs(
title = 'stat_dots(layout = "weave")',
subtitle = "aes(fill = after_stat(x < 6))"
) +
geom_vline(xintercept = 6, alpha = 0.25) +
scale_x_continuous(breaks = 2:10)
Sometimes you may want to include multiple different types of slabs in the same plot in order to take advantage of the features each slab type provides. For example, people often combine densities with dotplots to show the underlying datapoints that go into a density estimate, creating so-called rain cloud plots.
To use multiple slab geometries together, you can use the
side
parameter to change which side of the interval a slab
is drawn on and set the scale
parameter to something around
0.5
(by default it is 0.9
) so that the two
slabs do not overlap. We’ll also scale the halfeye slab thickness by
n
(the number of observations in each group) so that the
area of each slab represents sample size (and looks similar to the total
area of its corresponding dotplot).
We’ll use a subsample of of the data to show how it might look on a reasonably-sized dataset.
set.seed(12345) # for reproducibility
tibble(
abc = rep(c("a", "b", "b", "c"), 50),
value = rnorm(200, c(1, 8, 8, 3), c(1, 1.5, 1.5, 1))
) %>%
ggplot(aes(y = abc, x = value, fill = abc)) +
stat_slab(aes(thickness = after_stat(pdf*n)), scale = 0.7) +
stat_dotsinterval(side = "bottom", scale = 0.7, slab_linewidth = NA) +
scale_fill_brewer(palette = "Set2") +
ggtitle(
paste0(
'stat_slab(aes(thickness = after_stat(pdf*n)), scale = 0.7) +\n',
'stat_dotsinterval(side = "bottom", scale = 0.7, slab_linewidth = NA)'
),
'aes(fill = abc)'
)
A specialized variant of geom_dots()
,
geom_blur_dots()
, supports visualizing dotplots with blur
applied to each dot. stat_mcse_dots()
uses
geom_blur_dots()
with
posterior::mcse_quantile()
to show the error in each
quantile of a quantile dotplot:
increasing_samples %>%
ggplot(aes(x = x)) +
stat_mcse_dots(quantiles = 100) +
facet_wrap(~ n) +
labs(
title = "stat_mcse_dots(quantiles = 100)",
subtitle = "Monte Carlo Standard Error of each quantile shown as blur"
)
Custom blur functions can be selected using the blur
parameter, including the built-in blur_interval()
, which
draws an interval with a default width of 95%:
To demonstrate another useful plot type, the logit dotplot (courtesy Ladislas Nalborczyk), we’ll fit a logistic regression to some data on the petal length of the Iris versicolor and Iris virginica flowers.
First, we’ll demo varying the side
aesthetic to create
two dotplots that are “facing” each other:
scale_side_mirrored()
will set the side
aesthetic to "top"
or "bottom"
if two
categories are assigned to side
“. We also adjust the
scale
so that the dots don’t overlap:
iris_v = iris %>%
filter(Species != "setosa")
iris_v %>%
ggplot(aes(x = Petal.Length, y = Species, side = Species)) +
geom_dots(scale = 0.5) +
scale_side_mirrored(guide = "none") +
ggtitle(
"geom_dots(scale = 0.5)",
'aes(side = Species) + scale_side_mirrored()'
)
This can also be accomplished by setting side directly and omitting
scale_side_mirrored()
; e.g. via
aes(side = ifelse(Species == "virginica", "bottom", "top"))
.
Now we fit a logistic regression predicting species based on petal length:
##
## Call: glm(formula = Species == "virginica" ~ Petal.Length, family = binomial,
## data = iris_v)
##
## Coefficients:
## (Intercept) Petal.Length
## -43.781 9.002
##
## Degrees of Freedom: 99 Total (i.e. Null); 98 Residual
## Null Deviance: 138.6
## Residual Deviance: 33.43 AIC: 37.43
Then we can overlay a fit line as a stat_lineribbon()
(see vignette("lineribbon")
) on top of the mirrored
dotplots to create a logit dotplot:
# construct a prediction grid for the fit line
prediction_grid = with(iris_v,
data.frame(Petal.Length = seq(min(Petal.Length), max(Petal.Length), length.out = 100))
)
prediction_grid %>%
bind_cols(predict(m, ., se.fit = TRUE)) %>%
mutate(
# distribution describing uncertainty in log odds
log_odds = dist_normal(fit, se.fit),
# inverse-logit transform the log odds to get
# distribution describing uncertainty in Pr(Species == "virginica")
p_virginica = dist_transformed(log_odds, plogis, qlogis)
) %>%
ggplot(aes(x = Petal.Length)) +
geom_dots(
aes(y = as.numeric(Species == "virginica"), side = Species),
scale = 0.4,
data = iris_v
) +
stat_lineribbon(
aes(ydist = p_virginica), alpha = 1/4, fill = "#08306b"
) +
scale_side_mirrored(guide = "none") +
coord_cartesian(ylim = c(0, 1)) +
labs(
title = "logit dotplot: geom_dots() with stat_lineribbon()",
subtitle = 'aes(side = Species) + scale_side_mirrored()',
x = "Petal Length",
y = "Pr(Species = virginica)"
)