Returns a 3D model matrix for capture-recapture modeling in the form of a (giant) 2D matrix.
F.3d.model.matrix(formula, d1, d2)A formula object specifying covariates in a capture-recapture
model. Must not have a response, i.e., ~, followed by the names
of 2-D arrays or 1-D vectors contained inside calls to ivar or
tvar. See help(F.cjs.estim) for examples.
Magnitude of dimension 1 of the returned matrix. This is
always number of rows in the returned matrix. Usually, d1 = number
of individuals = number of rows in capture history matrix.
Magnitude of dimension 2 of the returned matrix. This is number of columns in the capture history matrix.
A (giant) 2-d matrix containing covariate values suitable for passing to
the Fortran code that does the estimation for MRA. This matrix has all the
2-d matrices of the model cbind-ed together. It's dimension is NAN x
NS*(number of coefficients). A convenient way to view the matrix is to assign
a 3-d dimension. I.e., if x is the result of a call to this function
and there are NX coefficients in the model,
then dim(x) <- c(NAN,NS,NX) makes a 3-d matrix with NAN rows, NS columns,
and NX pages. View the covariates for a single animal with x[3,,] or similar
statement.
Names of variables in the model are returned as attribute "variables". Whether the model has an intercept is returned as attribute "intercept".
This routine is intended to be called internally by the routines of MRA. General users should never have to call this routine.
This routine uses a call to eval with a model frame, and calls the
R internal model.matrix to
resolve the matrices in the formula. All matrices specified in the models
should be in the current scope and accessible to both eval and model.matrix.
See help(F.cjs.estim) for examples of ways to specify models.
# NOT RUN {
# Synthetic example with 10 animals and 5 occasions
nan <- 10
ns <- 5
sex <- as.factor(as.numeric(runif( nan ) > 0.5))
attr(sex,"ns") <- ns
x <- matrix( runif( nan*ns ) , nrow=nan, ncol=ns )
F.3d.model.matrix( ~ ivar(sex) + x, nan, ns )
# }
Run the code above in your browser using DataLab