Aids the eye in seeing patterns in the presence of overplotting.
geom_smooth
and stat_smooth
are effectively aliases: they
both use the same arguments. Use geom_smooth
unless you want to
display the results with a nonstandard geom.
geom_smooth(mapping = NULL, data = NULL, stat = "smooth", position = "identity", ..., method = "auto", formula = y ~ x, se = TRUE, na.rm = FALSE, show.legend = NA, inherit.aes = TRUE) stat_smooth(mapping = NULL, data = NULL, geom = "smooth", position = "identity", ..., method = "auto", formula = y ~ x, se = TRUE, n = 80, span = 0.75, fullrange = FALSE, level = 0.95, method.args = list(), 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 
position  Position adjustment, either as a string, or the result of a call to a position adjustment function. 
...  Other arguments passed on to 
method  Smoothing method (function) to use, eg. For If you have fewer than 1,000 observations but want to use the same 
formula  Formula to use in smoothing function, eg. 
se  Display confidence interval around smooth? (TRUE by default, see level to control.) 
na.rm  If 
show.legend  logical. Should this layer be included in the legends?

inherit.aes  If 
geom, stat  Use to override the default connection between

n  Number of points at which to evaluate smoother. 
span  Controls the amount of smoothing for the default loess smoother. Smaller numbers produce wigglier lines, larger numbers produce smoother lines. 
fullrange  Should the fit span the full range of the plot, or just the data? 
level  Level of confidence interval to use (0.95 by default). 
method.args  List of additional arguments passed on to the modelling
function defined by 
Calculation is performed by the (currently undocumented)
predictdf
generic and its methods. For most methods the standard
error bounds are computed using the predict()
method  the
exceptions are loess
which uses a tbased approximation, and
glm
where the normal confidence interval is constructed on the link
scale, and then backtransformed to the response scale.
geom_smooth
understands the following aesthetics (required aesthetics are in bold):
x
y
alpha
colour
fill
group
linetype
size
weight
ymax
ymin
Learn more about setting these aesthetics in vignette("ggplot2specs")
predicted value
lower pointwise confidence interval around the mean
upper pointwise confidence interval around the mean
standard error
See individual modelling functions for more details:
lm()
for linear smooths,
glm()
for generalised linear smooths,
loess()
for local smooths
#># Use span to control the "wiggliness" of the default loess smoother # The span is the fraction of points used to fit each local regression: # small numbers make a wigglier curve, larger numbers make a smoother curve. ggplot(mpg, aes(displ, hwy)) + geom_point() + geom_smooth(span = 0.3)#># Instead of a loess smooth, you can use any other modelling function: ggplot(mpg, aes(displ, hwy)) + geom_point() + geom_smooth(method = lm, se = FALSE)ggplot(mpg, aes(displ, hwy)) + geom_point() + geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE)# Smoothes are automatically fit to each group (defined by categorical # aesthetics or the group aesthetic) and for each facet ggplot(mpg, aes(displ, hwy, colour = class)) + geom_point() + geom_smooth(se = FALSE, method = lm)#>binomial_smooth < function(...) { geom_smooth(method = "glm", method.args = list(family = "binomial"), ...) } # To fit a logistic regression, you need to coerce the values to # a numeric vector lying between 0 and 1. ggplot(rpart::kyphosis, aes(Age, Kyphosis)) + geom_jitter(height = 0.05) + binomial_smooth()#> Warning: Computation failed in `stat_smooth()`: #> y values must be 0 <= y <= 1ggplot(rpart::kyphosis, aes(Age, as.numeric(Kyphosis)  1)) + geom_jitter(height = 0.05) + binomial_smooth()ggplot(rpart::kyphosis, aes(Age, as.numeric(Kyphosis)  1)) + geom_jitter(height = 0.05) + binomial_smooth(formula = y ~ splines::ns(x, 2))# But in this case, it's probably better to fit the model yourself # so you can exercise more control and see whether or not it's a good model