Revision 94dac762f5b1607c843aa97ea6a6caa68343fdb1 authored by Adrian Baddeley on 22 March 2019, 12:40:03 UTC, committed by cran-robot on 22 March 2019, 12:40:03 UTC
1 parent 80e2bf6
Concom.Rd
\name{Concom}
\alias{Concom}
\title{The Connected Component Process Model}
\description{
Creates an instance of the Connected Component point process model
which can then be fitted to point pattern data.
}
\usage{
Concom(r)
}
\arguments{
\item{r}{Threshold distance}
}
\value{
An object of class \code{"interact"}
describing the interpoint interaction
structure of the connected component process with disc radius \eqn{r}.
}
\details{
This function defines the interpoint interaction structure of a point
process called the connected component process.
It can be used to fit this model to point pattern data.
The function \code{\link{ppm}()}, which fits point process models to
point pattern data, requires an argument
of class \code{"interact"} describing the interpoint interaction
structure of the model to be fitted.
The appropriate description of the connected component interaction is
yielded by the function \code{Concom()}. See the examples below.
In \bold{standard form}, the connected component process
(Baddeley and \ifelse{latex}{\out{M\o ller}}{Moller}, 1989) with disc radius \eqn{r},
intensity parameter \eqn{\kappa}{\kappa} and interaction parameter
\eqn{\gamma}{\gamma} is a point process with probability density
\deqn{
f(x_1,\ldots,x_n) =
\alpha \kappa^{n(x)} \gamma^{-C(x)}
}{
f(x[1],\ldots,x[n]) =
\alpha . \kappa^n(x) . \gamma^(-C(x))
}
for a point pattern \eqn{x}, where
\eqn{x_1,\ldots,x_n}{x[1],\ldots,x[n]} represent the
points of the pattern, \eqn{n(x)} is the number of points in the
pattern, and \eqn{C(x)} is defined below.
Here \eqn{\alpha}{\alpha} is a normalising constant.
To define the term \code{C(x)}, suppose that we construct a planar
graph by drawing an edge between
each pair of points \eqn{x_i,x_j}{x[i],x[j]} which are less than
\eqn{r} units apart. Two points belong to the same connected component
of this graph if they are joined by a path in the graph.
Then \eqn{C(x)} is the number of connected components of the graph.
The interaction parameter \eqn{\gamma}{\gamma} can be any positive number.
If \eqn{\gamma = 1}{\gamma = 1} then the model reduces to a Poisson
process with intensity \eqn{\kappa}{\kappa}.
If \eqn{\gamma < 1}{\gamma < 1} then the process is regular,
while if \eqn{\gamma > 1}{\gamma > 1} the process is clustered.
Thus, a connected-component interaction process can be used to model either
clustered or regular point patterns.
In \pkg{spatstat}, the model is parametrised in a different form,
which is easier to interpret.
In \bold{canonical form}, the probability density is rewritten as
\deqn{
f(x_1,\ldots,x_n) =
\alpha \beta^{n(x)} \gamma^{-U(x)}
}{
f(x_1,\ldots,x_n) =
\alpha . \beta^n(x) \gamma^(-U(x))
}
where \eqn{\beta}{\beta} is the new intensity parameter and
\eqn{U(x) = C(x) - n(x)} is the interaction potential.
In this formulation, each isolated point of the pattern contributes a
factor \eqn{\beta}{\beta} to the probability density (so the
first order trend is \eqn{\beta}{\beta}). The quantity
\eqn{U(x)} is a true interaction potential, in the sense that
\eqn{U(x) = 0} if the point pattern \eqn{x} does not contain any
points that lie close together.
When a new point \eqn{u} is added to an existing point pattern
\eqn{x}, the rescaled potential \eqn{-U(x)} increases by
zero or a positive integer.
The increase is zero if \eqn{u} is not close to any point of \eqn{x}.
The increase is a positive integer \eqn{k} if there are
\eqn{k} different connected components of \eqn{x} that lie close to \eqn{u}.
Addition of the point
\eqn{u} contributes a factor \eqn{\beta \eta^\delta}{\beta * \eta^\delta}
to the probability density, where \eqn{\delta}{\delta} is the
increase in potential.
If desired, the original parameter \eqn{\kappa}{\kappa} can be recovered from
the canonical parameter by \eqn{\kappa = \beta\gamma}{\kappa = \beta * \gamma}.
The \emph{nonstationary} connected component process is similar except that
the contribution of each individual point \eqn{x_i}{x[i]}
is a function \eqn{\beta(x_i)}{\beta(x[i])}
of location, rather than a constant beta.
Note the only argument of \code{Concom()} is the threshold distance \code{r}.
When \code{r} is fixed, the model becomes an exponential family.
The canonical parameters \eqn{\log(\beta)}{log(\beta)}
and \eqn{\log(\gamma)}{log(\gamma)}
are estimated by \code{\link{ppm}()}, not fixed in
\code{Concom()}.
}
\seealso{
\code{\link{ppm}},
\code{\link{pairwise.family}},
\code{\link{ppm.object}}
}
\section{Edge correction}{
The interaction distance of this process is infinite.
There are no well-established procedures for edge correction
for fitting such models, and accordingly the model-fitting function
\code{\link{ppm}} will give an error message saying that the user must
specify an edge correction. A reasonable solution is
to use the border correction at the same distance \code{r}, as shown in the
Examples.
}
\examples{
# prints a sensible description of itself
Concom(r=0.1)
# Fit the stationary connected component process to redwood data
ppm(redwood, ~1, Concom(r=0.07), rbord=0.07)
# Fit the stationary connected component process to `cells' data
ppm(cells, ~1, Concom(r=0.06), rbord=0.06)
# eta=0 indicates hard core process.
# Fit a nonstationary connected component model
# with log-cubic polynomial trend
\dontrun{
ppm(swedishpines, ~polynom(x/10,y/10,3), Concom(r=7), rbord=7)
}
}
\references{
Baddeley, A.J. and \ifelse{latex}{\out{M\o ller}}{Moller}, J. (1989)
Nearest-neighbour Markov point processes and random sets.
\emph{International Statistical Review} \bold{57}, 89--121.
}
\author{
\spatstatAuthors
}
\keyword{spatial}
\keyword{models}
Computing file changes ...