Given a dataset with alluvial structure, stat_flow
calculates the centroids
(x
and y
) and heights (ymin
and ymax
) of the flows between each pair
of adjacent axes.
stat_flow( mapping = NULL, data = NULL, geom = "flow", position = "identity", decreasing = NULL, reverse = NULL, absolute = NULL, discern = FALSE, negate.strata = NULL, aes.bind = NULL, infer.label = FALSE, min.y = NULL, max.y = NULL, na.rm = FALSE, show.legend = NA, inherit.aes = TRUE, ... )
mapping  Set of aesthetic mappings created by 

data  The data to be displayed in this layer. There are three options: If A A 
geom  The geometric object to use display the data; override the default. 
position  Position adjustment, either as a string, or the result of a call to a position adjustment function. 
decreasing  Logical; whether to arrange the strata at each axis
in the order of the variable values ( 
reverse  Logical; if 
absolute  Logical; if some cases or strata are negative,
whether to arrange them (respecting 
discern  Passed to 
negate.strata  A vector of values of the 
aes.bind  At what grouping level, if any, to prioritize differentiation
aesthetics when ordering the lodes within each stratum. Defaults to

infer.label  Logical; whether to assign the 
min.y  Numeric; bounds on the heights of the strata to be
rendered. Use these bounds to exclude strata outside a certain range, for
example when labeling strata using 
max.y  Numeric; bounds on the heights of the strata to be
rendered. Use these bounds to exclude strata outside a certain range, for
example when labeling strata using 
na.rm  Logical:
if 
show.legend  logical. Should this layer be included in the legends?

inherit.aes  If 
...  Additional arguments passed to 
stat_alluvium
, stat_flow
, and stat_stratum
require one
of two sets of aesthetics:
x
and at least one of alluvium
and stratum
any number of axis[09]*
(axis1
, axis2
, etc.)
Use x
, alluvium
, and/or stratum
for data in lodes format
and axis[09]*
for data in alluvia format (see alluvialdata
).
Arguments to parameters inconsistent with the format will be ignored.
Additionally, each stat_*()
accepts the following optional
aesthetics:
y
weight
order
group
label
y
controls the heights of the alluvia,
and may be aggregated across equivalent observations.
weight
applies to the computed variables (see that section below)
but does not affect the positional aesthetics.
order
, recognized by stat_alluvium()
and stat_flow()
, is used to
arrange the lodes within each stratum. It tolerates duplicates and takes
precedence over the differentiation aesthetics (when aes.bind
is not
"none"
) and lode guidance with respect to the remaining axes. (It replaces
the deprecated parameter lode.ordering
.)
group
is used internally; arguments are ignored.
label
is used to label the strata or lodes and must take a unique value
across the observations within each stratum or lode.
These and any other aesthetics are aggregated as follows:
Numeric aesthetics, including y
, are summed.
Character and factor aesthetics, including label
,
are assigned to strata or lodes provided they take unique values across the
observations within each (and are otherwise assigned NA
).
These can be used with
ggplot2::after_stat()
to control aesthetic evaluation.
n
number of cases in lode
count
cumulative weight of lode
prop
weighted proportion of lode
stratum
value of variable used to define strata
deposit
order in which (signed) strata are deposited
lode
lode label distilled from alluvia
(stat_alluvium()
and stat_flow()
only)
flow
direction of flow "to"
or "from"
from its axis
(stat_flow()
only)
The numerical variables n
, count
, and prop
are calculated after the
data are grouped by x
and weighted by weight
(in addition to y
).
The integer variable deposit
is used internally to sort the data before
calculating heights. The character variable lode
is obtained from
alluvium
according to distill
.
stat_stratum
, stat_alluvium
, and stat_flow
order strata and lodes
according to the values of several parameters, which must be held fixed
across every layer in an alluvial plot. These packagespecific options set
global values for these parameters that will be defaulted to when not
manually set:
ggalluvial.decreasing
(each stat_*
): defaults to NA
.
ggalluvial.reverse
(each stat_*
): defaults to TRUE
.
ggalluvial.absolute
(each stat_*
): defaults to TRUE
.
ggalluvial.cement.alluvia
(stat_alluvium
): defaults to FALSE
.
ggalluvial.lode.guidance
(stat_alluvium
): defaults to "zigzag"
.
ggalluvial.aes.bind
(stat_alluvium
and stat_flow
): defaults to
"none"
.
See base::options()
for how to use options.
The previously defunct parameters weight
and aggregate.wts
have been
discontinued. Use y
and cement.alluvia
instead.
ggplot2::layer()
for additional arguments and
geom_alluvium()
and
geom_flow()
for the corresponding geoms.
Other alluvial stat layers:
stat_alluvium()
,
stat_stratum()
# illustrate positioning ggplot(as.data.frame(Titanic), aes(y = Freq, axis1 = Class, axis2 = Sex, axis3 = Age, color = Survived)) + stat_stratum(geom = "errorbar") + geom_line(stat = "flow") + stat_flow(geom = "pointrange") + geom_text(stat = "stratum", aes(label = after_stat(stratum))) + scale_x_discrete(limits = c("Class", "Sex", "Age")) # alluviumflow comparison data(vaccinations) gg < ggplot(vaccinations, aes(x = survey, stratum = response, alluvium = subject, y = freq, fill = response)) + geom_stratum(alpha = .5) + geom_text(aes(label = response), stat = "stratum") # rightward alluvial aesthetics for vaccine survey data gg + geom_flow(stat = "alluvium", lode.guidance = "forward") # memoryless flows for vaccine survey data gg + geom_flow() # size filter examples gg < ggplot(vaccinations, aes(y = freq, x = survey, stratum = response, alluvium = subject, fill = response, label = response)) + stat_stratum(alpha = .5) + geom_text(stat = "stratum") # omit small flows gg + geom_flow(min.y = 50) # omit large flows gg + geom_flow(max.y = 100) # negate missing entries ggplot(vaccinations, aes(y = freq, x = survey, stratum = response, alluvium = subject, fill = response, label = response, alpha = response != "Missing")) + stat_stratum(negate.strata = "Missing") + geom_flow(negate.strata = "Missing") + geom_text(stat = "stratum", alpha = 1, negate.strata = "Missing") + scale_alpha_discrete(range = c(.2, .6)) + guides(alpha = "none") #> Warning: Using alpha for a discrete variable is not advised. # aesthetics that vary betwween and within strata data(vaccinations) vaccinations$subgroup < LETTERS[1:2][rbinom( n = length(unique(vaccinations$subject)), size = 1, prob = .5 ) + 1][vaccinations$subject] ggplot(vaccinations, aes(x = survey, stratum = response, alluvium = subject, y = freq, fill = response, label = response)) + geom_flow(aes(alpha = subgroup)) + scale_alpha_discrete(range = c(1/3, 2/3)) + geom_stratum(alpha = .5) + geom_text(stat = "stratum") #> Warning: Using alpha for a discrete variable is not advised. # can even set aesthetics that vary both ways ggplot(vaccinations, aes(x = survey, stratum = response, alluvium = subject, y = freq, label = response)) + geom_flow(aes(fill = interaction(response, subgroup)), aes.bind = "flows") + scale_alpha_discrete(range = c(1/3, 2/3)) + geom_stratum(alpha = .5) + geom_text(stat = "stratum") #> Warning: Using alpha for a discrete variable is not advised.