Class "spatial.neighbor"

DESCRIPTION:

Class of objects used to define neighbor relationships for spatial data on a regular or irregular lattice.

GENERATION:

This class of objects is constructed using the function spatial.neighbor. Alternatively, the functions read.neighbor, or neighbor.grid may be used. In general, the user must construct these objects whenever estimates of spatial correlation and spatial regression are desired.

An object of class "spatial.neighbor" contains all the information required to determine which spatial units on a region of interest are neighbors, as well as the strength of their relationship.

METHODS:

The class "spatial.neighbor" has associated methods, print.spatial.neighbor , plot.spatial.neighbor, and summary.spatial.neighbor .

INHERITANCE:

Class "spatial.neighbor" inherits from class "data.frame".

STRUCTURE:

The "spatial.neighbor" object is in essence a data frame with additional attributes. Each row of the data frame denotes a pair of neighboring spatial units. The data frame contains the following columns:

VALUE:

row.id
the row index in the neighbor matrix that corresponds to a region or spatial unit. This implies a numbering of regions from 1 to the total number of regions.
col.id
the column index in the neighbor matrix that corresponds to the neighbor of the region defined by the corresponding element of row.id.
weights
a numeric value giving the relative strength of the neighbor relationship. The larger the value, the stronger the relationship.
matrix
if multiple types of neighbor matrices are possible, this column contains the type of the neighbor this weight represents - it gives a numeric identifier for each spatial neighbor [contiguity] matrix.

SPECIAL ATTRIBUTES:

VALUE:

nregion
the number of total regions in the study. The row and column identifiers given in row.id and col.id might not include ALL the spatial units in the area of interest. This happens when units are isolated, i.e. have no neighboring regions. In this case, nregion must be used to determine the total number of rows and columns in the contiguity matrix.
symmetric
logical flag. It provides an indication of whether the contiguity matrix is symmetric ( TRUE) or not ( FALSE). If TRUE, only the weights for the upper (or lower) triangle of the contiguity matrix need to be specified in the object. Use the function spatial.weights to expand the full symmetric weights matrix.

DETAILS:

An object of class "spatial.neighbor" is a sparse matrix representation of a square matrix (or a number of square matrices).

The function plot.spatial.neighbor will show a graphical view of the spatial.neighbor object and summary.spatial.neighbor will compute summary statistics on the object.

The functions spatial.multiply, and spatial.cg.solve can be used to form products of the form rho[i]*N[i]*x and (rho[i]*N[i])^(-1)*x, for neighbor weight matrices N[i], vector of constants or parameters, rho[i] , and arbitrary vectors x, should that be needed to form a neighbor or contiguity matrix as a weighted linear combination of others.

SEE ALSO:

, , , , , , .