1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
|
\name{densityPlot}
\alias{densityPlot}
\alias{densityPlot.default}
\alias{densityPlot.formula}
\alias{adaptiveKernel}
\alias{depan}
\alias{dbiwt}
\title{
Nonparametric Density Estimates
}
\description{
\code{densityPlot} contructs and graphs nonparametric density estimates, possibly conditioned on a factor, using the standard \R{} \code{\link{density}} function or by default \code{adaptiveKernel}, which computes an adaptive kernel density estimate.
\code{depan} provides the Epanechnikov kernel and \code{dbiwt} provides the biweight kernel.
}
\usage{
densityPlot(x, ...)
\method{densityPlot}{default}(x, g, method=c("adaptive", "kernel"),
bw=if (method == "adaptive") bw.nrd0 else "SJ", adjust=1,
kernel, xlim, ylim,
normalize=FALSE, xlab=deparse(substitute(x)), ylab="Density", main="",
col=carPalette(), lty=seq_along(col), lwd=2, grid=TRUE,
legend=TRUE, show.bw=FALSE, rug=TRUE, ...)
\method{densityPlot}{formula}(formula, data=NULL, subset,
na.action=NULL, xlab, ylab, main="", legend=TRUE, ...)
adaptiveKernel(x, kernel=dnorm, bw=bw.nrd0, adjust=1.0, n=500,
from, to, cut=3, na.rm=TRUE)
depan(x)
dbiwt(x)
}
\arguments{
\item{x}{a numeric variable, the density of which is estimated; for
\code{depan} and \code{dbiwt}, the argument of the kernel function.}
\item{g}{an optional factor to divide the data.}
\item{formula}{an \R{} model formula, of the form \code{~ variable} to estimate the unconditional
density of \code{variable}, or \code{variable ~ factor} to estimate the density of \code{variable}
within each level of \code{factor}.}
\item{data}{an optional data frame containing the data.}
\item{subset}{an optional vector defining a subset of the data.}
\item{na.action}{a function to handle missing values; defaults to the value of the \R{} \code{na.action} option,
initially set to \code{\link{na.omit}}.}
\item{method}{either \code{"adaptive"} (the default) for an adaptive-kernel estimate or \code{"kernel"} for a fixed-bandwidth kernel estimate.}
\item{bw}{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 \code{\link{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, \code{bw}
may be a vector of values, one for each group. See \code{\link{density}} and \code{\link{bw.SJ}} for details of the kernel estimator.}
\item{adjust}{a multiplicative adjustment factor for the bandwidth; the default, \code{1}, indicates no adjustment;
if plotting by groups, \code{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 \code{adjust} < 1, particularly for the adaptive-kernel estimator.}
\item{kernel}{for \code{densityPlot} this is the name of the kernel function for the kernel estimator (the default is \code{"gaussian"}, see \code{\link{density}});
or a kernel function for the adaptive-kernel estimator (the default is \code{dnorm}, producing the Gaussian kernel).
For \code{adaptivekernel} this is a kernel function, defaulting to \code{dnorm}, which is the Gaussian kernel (standard-normal density).}
\item{xlim, ylim}{axis limits; if missing, determined from the range of x-values at which the densities are estimated and the estimated densities.}
\item{normalize}{if \code{TRUE} (the default is \code{FALSE}), the estimated densities are rescaled to integrate approximately to 1; particularly useful if the
density is estimated over a restricted domain, as when \code{from} or \code{to} are specified.}
\item{xlab}{label for the horizontal-axis; defaults to the name of the variable \code{x}.}
\item{ylab}{label for the vertical axis; defaults to \code{"Density"}.}
\item{main}{plot title; default is empty.}
\item{col}{vector of colors for the density estimate(s); defaults to the color \code{\link{carPalette}}.}
\item{lty}{vector of line types for the density estimate(s); defaults to the successive integers, starting at 1.}
\item{lwd}{line width for the density estimate(s); defaults to 2.}
\item{grid}{if \code{TRUE} (the default), grid lines are drawn on the plot.}
\item{legend}{a list of up to two named elements: \code{location}, for the legend when densities are plotted for several groups, defaults to
\code{"upperright"} (see \code{\link{legend}}); and \code{title} of the legend, which defaults to the name of the grouping factor. If \code{TRUE},
the default, the default values are used; if \code{FALSE}, the legend is suppressed.}
\item{n}{number of equally spaced points at which the adaptive-kernel estimator is evaluated; the default is \code{500}.}
\item{from, to, cut}{the range over which the density estimate is computed; the default, if missing, is \code{min(x) - cut*bw, max(x) + cut*bw}.}
\item{na.rm}{remove missing values from \code{x} in computing the adaptive-kernel estimate? The default is \code{TRUE}.}
\item{show.bw}{if \code{TRUE}, show the bandwidth(s) in the horizontal-axis label or (for multiple groups)
the legend; the default is \code{FALSE}.}
\item{rug}{if \code{TRUE} (the default), draw a rug plot (one-dimentional scatterplot) at the bottom of the density estimate.}
\item{\dots}{arguments to be passed down to graphics functions.}
}
\details{
If you use a different kernel function than the default \code{dnorm} that has a
standard deviation different from 1 along with an automatic rule
like the default function \code{bw.nrd0}, you can attach an attribute to the kernel
function named \code{"scale"} that gives its standard deviation. This is true for
the two supplied kernels, \code{depan} and \code{dbiwt}
}
\value{
\code{densityPlot} invisibly returns the \code{"density"} object computed (or list of \code{"density"} objects) and draws a graph.
\code{adaptiveKernel} returns an object of class \code{"density"}
(see \code{\link{density})}.
}
\references{
Fox, J. and Weisberg, S. (2019)
\emph{An R Companion to Applied Regression}, Third Edition, Sage.
W. N. Venables and B. D. Ripley (2002) \emph{Modern Applied Statistics with S}. New York: Springer.
B.W. Silverman (1986) \emph{Density Estimation for Statistics and Data Analysis}. London: Chapman and Hall.
}
\author{
John Fox \email{jfox@mcmaster.ca}
}
\seealso{
\code{\link{density}}, \code{\link{bw.SJ}}, \code{\link{plot.density}}
}
\examples{
densityPlot(~ income, show.bw=TRUE, method="kernel", data=Prestige)
densityPlot(~ income, show.bw=TRUE, data=Prestige)
densityPlot(~ income, from=0, normalize=TRUE, show.bw=TRUE, data=Prestige)
densityPlot(income ~ type, data=Prestige)
densityPlot(~ income, show.bw=TRUE, method="kernel", data=Prestige)
densityPlot(~ income, show.bw=TRUE, data=Prestige)
densityPlot(~ income, from=0, normalize=TRUE, show.bw=TRUE, data=Prestige)
densityPlot(income ~ type, kernel=depan, data=Prestige)
densityPlot(income ~ type, kernel=depan, legend=list(location="top"), data=Prestige)
plot(adaptiveKernel(UN$infantMortality, from=0, adjust=0.75), col="magenta")
lines(density(na.omit(UN$infantMortality), from=0, adjust=0.75), col="blue")
rug(UN$infantMortality, col="cyan")
legend("topright", col=c("magenta", "blue"), lty=1,
legend=c("adaptive kernel", "kernel"), inset=0.02)
plot(adaptiveKernel(UN$infantMortality, from=0, adjust=0.75), col="magenta")
lines(density(na.omit(UN$infantMortality), from=0, adjust=0.75), col="blue")
rug(UN$infantMortality, col="cyan")
legend("topright", col=c("magenta", "blue"), lty=1,
legend=c("adaptive kernel", "kernel"), inset=0.02)
}
\keyword{hplot}
|
