stat_summary(mapping = NULL, data = NULL, geom = "pointrange", position = "identity", ...)
aes
or aes_string
. Only
needs to be set at the layer level if you are overriding
the plot defaults.layer
. This can include aesthetics whose
values you want to set, not map. See layer
for more details.a data.frame with additional columns: fun.dataComplete summary function. Should take data frame as input and return data frame as output fun.yminymin summary function (should take numeric vector and return single number) fun.yy summary function (should take numeric vector and return single number) fun.ymaxymax summary function (should take numeric vector and return single number)
stat_summary
allows for tremendous flexibilty in
the specification of summary functions. The summary
function can either operate on a data frame (with
argument name fun.data
) or on a vector
(fun.y
, fun.ymax
, fun.ymin
).
A simple vector function is easiest to work with as you can return a single number, but is somewhat less flexible. If your summary function operates on a data.frame it should return a data frame with variables that the geom can use.
stat_summary
understands the following aesthetics (required aesthetics are in bold):
x
y
# Basic operation on a small dataset d <- qplot(cyl, mpg, data=mtcars) d + stat_summary(fun.data = "mean_cl_boot", colour = "red")p <- qplot(cyl, mpg, data = mtcars, stat="summary", fun.y = "mean") p# Don't use ylim to zoom into a summary plot - this throws the # data away p + ylim(15, 30)Warning message: Removed 9 rows containing missing values (stat_summary).# Instead use coord_cartesian p + coord_cartesian(ylim = c(15, 30))# You can supply individual functions to summarise the value at # each x: stat_sum_single <- function(fun, geom="point", ...) { stat_summary(fun.y=fun, colour="red", geom=geom, size = 3, ...) } d + stat_sum_single(mean)d + stat_sum_single(mean, geom="line")d + stat_sum_single(median)d + stat_sum_single(sd)d + stat_summary(fun.y = mean, fun.ymin = min, fun.ymax = max, colour = "red")d + aes(colour = factor(vs)) + stat_summary(fun.y = mean, geom="line")# Alternatively, you can supply a function that operates on a data.frame. # A set of useful summary functions is provided from the Hmisc package: stat_sum_df <- function(fun, geom="crossbar", ...) { stat_summary(fun.data=fun, colour="red", geom=geom, width=0.2, ...) } d + stat_sum_df("mean_cl_boot")d + stat_sum_df("mean_sdl")d + stat_sum_df("mean_sdl", mult=1)d + stat_sum_df("median_hilow")# There are lots of different geoms you can use to display the summaries d + stat_sum_df("mean_cl_normal")d + stat_sum_df("mean_cl_normal", geom = "errorbar")d + stat_sum_df("mean_cl_normal", geom = "pointrange")d + stat_sum_df("mean_cl_normal", geom = "smooth")# Summaries are more useful with a bigger data set: mpg2 <- subset(mpg, cyl != 5L) m <- ggplot(mpg2, aes(x=cyl, y=hwy)) + geom_point() + stat_summary(fun.data = "mean_sdl", geom = "linerange", colour = "red", size = 2, mult = 1) + xlab("cyl") m# An example with highly skewed distributions: set.seed(596) mov <- movies[sample(nrow(movies), 1000), ] m2 <- ggplot(mov, aes(x= factor(round(rating)), y=votes)) + geom_point() m2 <- m2 + stat_summary(fun.data = "mean_cl_boot", geom = "crossbar", colour = "red", width = 0.3) + xlab("rating") m2# Notice how the overplotting skews off visual perception of the mean # supplementing the raw data with summary statistics is _very_ important # Next, we'll look at votes on a log scale. # Transforming the scale means the data are transformed # first, after which statistics are computed: m2 + scale_y_log10()# Transforming the coordinate system occurs after the # statistic has been computed. This means we're calculating the summary on the raw data # and stretching the geoms onto the log scale. Compare the widths of the # standard errors. m2 + coord_trans(y="log10")
geom_errorbar
,
geom_pointrange
,
geom_linerange
, geom_crossbar
for geoms to display summarised data