| SGCS-funs {SGCS} | R Documentation |
Statistical measures based on edges of a geometric graph structure over a given point pattern data.
| Date: | 2009-03-04 |
| License: | GPL v2 or later |
The graph structure is in this version geometric graph, meaning points x and y are connected
if ||x-y||<R for the given range parameter R. Notice that in clustering function, the R equals to
the parameter r.
The main function is art1Fun, but the use of shortcuts confun, cumconfun and clustfun is highly
encouraged.
confun(pp, R=NULL, h=NULL, ...)
cumconfun(pp, R=NULL, ...)
clustfun(pp, ...)
art1Fun(pp, r = NULL, funtype = 1, fpar = 0, toroidal = FALSE, dbg = FALSE,
doDists = FALSE, doWeights = FALSE, prepR = 0, relative = FALSE,
prepGraph = NULL, minusR = NULL, included = NULL, parvec = NULL)
pp |
All: Point pattern object. ppp-objects from package spatstat are safe. |
R |
confun, cumconfun: Connection radius of the graph. If NULL, R=1/sqrt(lambda). |
h |
confun: Smoothing parameter in kernel-estimation. Box kernel width = 2h. |
... |
Parametes for the function art1Fun: |
r |
Vector of the range parameters in which to calculate the function value. if NULL, a range from 0 to 1/3 of window length is used. |
funtype |
(shortcuts handle) Which function to calculate. confun=1, cumconfun=2, clustfun=3. |
fpar |
(shortcuts handle) Additional function parameter(s): confun fpar=c(R,h), cumconfun fpar=R. |
toroidal |
Toroidal correction of rectangular window. |
dbg |
Print additional messages. |
doDists |
Precalculate distances for faster computation. Be aware of memory consumption O(n^2). |
doWeights |
Precalculate confun translation correction weights for faster computation. Be aware of memory consumption O(n^2). |
prepR |
Preparation distance: If >0, precalculate a graph, from which it is easier to find needed connections. Useful with other graphs (in the future). |
relative |
Scale the range parameter to a unit range, i.e. scale r with sqrt(pi*lambda). Good when comparing different phenomena in different spatial scales. |
prepGraph |
An optional graph-object from spatgraphs to be used as basis for finding connections between points. Good for big datasets. |
minusR |
Minus-correction parameter. Rectangular window required, NULL means no correction. |
included |
Point pattern size boolean vector for inclusion of points in computation. NULL means all points. |
parvec |
Same as r. |
An object of class graphfcl, which can be plotted with the generic plot-command. Contains the given parameters and
calculated function values, and an simulation based estimation of Poisson values. Note that the Poisson values (and the limits drawn
if flag limits=TRUE) is correct only when the R=1/sqrt(lambda).
Tuomas Rajala
University of Jyvaskyla, Finland
tarajala@maths.jyu.fi
Rajala, Penttinen: Spatial clustering and graph feature statistics (working title).