either formula providing a single dependent variable (y) and
an single independent variable (x) to use as coordinates in the
scatter plot or a numeric vector of x locations
y
numeric vector of y locations
data
an optional data.frame, list, or environment contianing
the variables used in the model (and in subset). If not found in
data, the variables are taken from environment(formula),
typically the environment from which lm is called.
subset
an optional vector specifying a subset of observations to be
used in the fitting process.
na.action
a function which indicates what should happen when
the data contain NAs. The default is set by the na.action
setting of options, and is na.fail if that is unset. The
factory-fresh default is na.omit. Another possible value is
NULL, no action. Value na.exclude can be useful.
...
Additional plotting parameters
xlab, ylab
x and y axis labels
add
Boolean indicating whether the local mean and standard
deviation lines should be added to an existing plot. Defaults to
FALSE.
sd
Vector of multiples of the standard devation that should be
plotted. 0 gives the mean, -1 gives the mean minus
one standard deviation, etc. Defaults to -2:2.
sd.col,sd.lwd,sd.lty
Color, line width, and line type of each plotted line.
method, width, n
Parameters controlling the smoothing. See the
help page for wapply for details.
Details
bandplot was created to look for changes in the mean or
variance of scatter plots, particularly plots of regression residuals.
The local mean and standard deviation are calculated by calling
'wapply'. By default, bandplot asks wapply to smooth using intervals
that include the nearest 1/5 of the data. See the documentation of
that function for details on the algorithm.
Value
Invisibly returns a list containing the x,y points plotted for each line.
# fixed mean, changing variance
x <- 1:1000
y <- rnorm(1000, mean=1, sd=1 + x/1000 )
bandplot(x,y)
bandplot(y~x)
# fixed varance, changing mean
x <- 1:1000
y <- rnorm(1000, mean=x/1000, sd=1)
bandplot(x,y)
#
# changing mean and variance
#
x <- abs(rnorm(500))
y <- rnorm(500, mean=2*x, sd=2+2*x)
# the changing mean and dispersion are hard to see whith the points alone:
plot(x,y )
# regression picks up the mean trend, but not the change in variance
reg <- lm(y~x)
summary(reg)
abline(reg=reg, col="blue", lwd=2)
# using bandplot on the original data helps to show the mean and
# variance trend
bandplot(y ~ x)
# using bandplot on the residuals helps to see that regression removes
# the mean trend but leaves the trend in variability
bandplot(predict(reg),resid(reg))