spatial.neighbor (and an optional set of
neighbor coefficients),
spatial.weights computes the spatial weight
matrix.
spatial.weights(neighbor, parameters=NULL, region.id=NULL)
spatial.neighbor used (in conjunction with the
elements of argument
parameters) to compute a the spatial weight
matrix. In situations with more than one weight matrix (see routine
spatial.neighbor, and the DETAILS section below),
spatial.weights
computes a weighted sum of the weight matrices.
matrix of argument
neighbor.
If
parameters is not specified, a vector of all
1s is assumed.
row.id and
col.id of argument
neighbor are
not integer valued variables with sequential values from 1 to the
number or regions in the lattice, then argument
region.id must be
specified and is used to obtain a sequential coding of the lattice
regions.
A[i] denote the spatial weight matrix of type
i, and let
parameters[i]
denote its corresponding parameter value. Then
spatial.weights
returns the linear sum over
i of
parameter[i]*A[i].
Argument
neighbor is a sparse matrix representation of one or more
matrices used to indicate the strength of neighbor relationships
between regions on a regular or irregular spatial lattice.
Let
a[i,j] denote a non-zero element of the k-th weight matrix.
This implies that regions
i and
j are neighbors of the k-th kind,
and the strength
of this relationship is given by the magnitude of
a[i,j].
spatial.weights can compute a weighted sum of the matrices
A, where
the weights in the linear combination are given by the elements in
argument
parameters. In practice, you might want to use routine
spatial.weights
when, for example, you want to compute the spatial
covariance matrix for a spatial process.
row.id <- c(1,1,2,3) col.id <- c(2,3,3,4) neighbor <- spatial.neighbor(row.id=row.id, col.id=col.id, symmetric=T) spatial.weights(neighbor)