spp
.
intensity(object, method = "kernel", bw, boundary = bbox(object), nx, ny, edge = F, ...)
"spp"
.
"kernel"
and
"gauss2d"
.
This should be a numeric value. The same value is used
each of the
x
or the
y
directions. This value needs to be determined
by the user mostly by trial-and-error. A good starting value might be
1/4
of the diameter of the boundary region.
object
.
This must be one of
"basic"
,
"kernel"
,
"binning"
, or
"gauss2d"
.
See the DETAILS section for each definition. Defaults to
"kernel"
which
applies a quartic kernel to the two-dimensional process.
Partial string matching is allowed.
boundary
should be
given as a list with named components
"x"
and
"y"
denoting the
corners of the rectangular region. For example, for the unit square
the boundary could be given as
bbox(x=c(0,1),y=c(0,1))
, the bounding
box of two diagonally opposed points.
Defaults to a rectangle covering the range of all the points in
object
.
object
.
object
.
object
so it may slow down computation.
Default is
FALSE
.
loess
smoother. In
particular:
loess
smoothing parameter, used by
method ="binning"
to smooth the
bin counts. See the
loess
help file for more information. Defaults to
0.75
with higher values producing smoother estimates.
if
method="basic"
a number is returned. In all other cases
the return value is a list with 4 components as follows:
x
,
y
respectively.
"call"
describing the
call that generated the resulting object.
The returned value may be given to the S-PLUS function
image
for
a graphical display of intensity over the area enclosed by
boundary
.
When
method="binning"
, the points are binned into a
nx
by
ny
grid
of counts and these are smoothed out using a call to the S-PLUS function
loess
.
When
method="kernel"
, a quartic kernel is placed at each cross
point on an overlaid
nx
by
ny
grid and a two-dimensional kernel
smoothing operation performed. This kernel is computed as the
product of two univariate quartic kernels, and so it is not rotationally
invariant. See the SPATIALSTATS function
kern2d
for
more information on the formulae used for these computations.
When
method ="gauss2d"
, a gaussian kernel is used instead of the
quartic of above.
When
edge
is set to
TRUE
, a lot more memory will be required.
Increasing
options("object.size")
might be necessary.
Diggle, Peter J. (1983).
Statistical Analysis of Spatial Point Patterns.
Academic Press, London.
Venables, W. N. and Ripley, B. D. (1994).
Modern Applied Statistics with S-Plus.
Springer-Verlag, New York.
image(intensity(lansing, "binning", nx=50, ny=50, span=.1)) image(intensity(lansing, bw=.3))