Nonparametric Density Estimates

densityPlot contructs and graphs nonparametric density estimates, possibly conditioned on a factor. By default it uses the standard R density function or optionally adaptiveKernel.

densityPlot(x, ...)

# S3 method for default densityPlot(x, g, method=c("kernel", "adaptive"), bw=if (method == "adaptive") bw.nrd0 else "SJ", adjust=1, kernel, xlim, ylim, normalize=FALSE, xlab=deparse(substitute(x)), ylab="Density", col=palette(), lty=seq_along(col), lwd=2, grid=TRUE, legend.location="topright", legend.title=deparse(substitute(g)),, rug=TRUE, ...) # S3 method for formula densityPlot(formula, data = NULL, subset, na.action = NULL, xlab, ylab, ...)

adaptiveKernel(x, kernel=dnorm, bw=bw.nrd0, adjust=1.0, n=500, from, to, cut=3, na.rm=TRUE)


a numeric variable, the density of which is estimated.


an optional factor to divide the data.


an R model formula, of the form ~ variable to estimate the unconditional density of variable, or variable ~ factor to estimate the density of variable within each level of factor.


an optional data frame containing the data.


an optional vector defining a subset of the data.


a function to handle missing values; defaults to the value of the R na.action option, initially set to na.omit.


either "adaptive" (the default) for an adaptive-kernel estimate or "kernel" for a fixed-bandwidth kernel estimate.


the geometric mean bandwidth for the adaptive-kernel or bandwidth of the kernel density estimate(s). Must be a numerical value or a function to compute the bandwidth (default bw.nrd0) for the adaptive kernel estimate; for the kernel estimate, may either the quoted name of a rule to compute the bandwidth, or a numeric value. If plotting by groups, bw may be a vector of values, one for each group. See density and bw.SJ for details of the kernel estimator.


a multiplicative adjustment factor for the bandwidth; the default, 1, indicates no adjustment; if plotting by groups, adjust may be a vector of adjustment factors, one for each group. The default bandwidth-selection rule tends to give a value that's too large if the distribution is asymmetric or has multiple modes; try setting adjust < 1, particularly for the adaptive-kernel estimator.


for densityPlot this is the name of the kernel function for the kernel estimator (the default is "gaussian", see density); or a kernel function for the adaptive-kernel estimator (the default is dnorm, producing the Gaussian kernel). For adaptivekernel this is a kernel function, defaulting to dnorm, which is the Gaussian kernel (standard-normal density).

xlim, ylim

axis limits; if missing, determined from the range of x-values at which the densities are estimated and the estimated densities.


if TRUE (the default is FALSE), the estimated densities are rescaled to integrate approximately to 1; particularly useful if the density is estimated over a restricted domain, as when from or to are specified.


label for the horizontal-axis; defaults to the name of the variable x.


label for the vertical axis; defaults to "Density".


vector of colors for the density estimate(s); defaults to the color palette.


vector of line types for the density estimate(s); defaults to the successive integers, starting at 1.


line width for the density estimate(s); defaults to 2.


if TRUE (the default), grid lines are drawn on the plot.


location for the legend when densities are plotted for several groups; defaults to "upperright"; see legend.


label for the legend, which is drawn if densities are plotted by groups; the default is the name of the factor g.


number of equally spaced points at which the adaptive-kernel estimator is evaluated; the default is 500.

from, to, cut

the range over which the density estimate is computed; the default, if missing, is min(x) - cut*bw, max(x) + cut*bw.


remove missing values from x in computing the adaptive-kernel estimate? The default is TRUE.

if TRUE, show the bandwidth(s) in the horizontal-axis label or (for multiple groups) the legend; the default is FALSE.


if TRUE (the default), draw a rug plot (one-dimentional scatterplot) at the bottom of the density estimate.

arguments to be passed down.


densityPlot invisibly returns the "density" object computed (or list of "density" objects) and draws a graph. adaptiveKernel returns an object of class "density" (see density).


W. N. Venables and B. D. Ripley (2002) Modern Applied Statistics with S. New York: Springer.

B.W. Silverman (1986) Density Estimation for Statistics and Data Analysis. London: Chapman and Hall.

See Also

density, bw.SJ, plot.density

  • densityPlot
  • densityPlot.default
  • densityPlot.formula
  • adaptiveKernel
densityPlot(~ income,, data=Prestige)
densityPlot(~ income, method="adaptive",, data=Prestige)
densityPlot(~ income, method="adaptive", from=0, normalize=TRUE,, data=Prestige)

densityPlot(income ~ type, method="adaptive", data=Prestige)

plot(adaptiveKernel(UN$infant.mortality, from=0, adjust=0.75), col="magenta")
lines(density(na.omit(UN$infant.mortality), from=0, adjust=0.75), col="blue")
rug(UN$infant.mortality, col="cyan")
legend("topright", col=c("magenta", "blue"), lty=1, 
  legend=c("adaptive kernel", "kernel"), inset=0.02)
# }
Documentation reproduced from package car, version 2.1-6, License: GPL (>= 2)

Community examples

Looks like there are no examples yet.