secr.design.MS: Construct Detection Model Design Matrices and Lookups

Description

Internal functions used by secr.fit.

Usage

secr.design.MS (capthist, models, timecov = NULL, sessioncov = NULL, 
    groups = NULL, hcov = NULL, dframe = NULL, naive = FALSE, CL = FALSE,
    keep.dframe = FALSE, full.dframe = FALSE, ignoreusage = FALSE, contrasts = NULL, ...)
make.lookup (tempmat)
insertdim (x, dimx, dims)

Value

For secr.design.MS, a list with the components

designMatrices: list of reduced design matrices, one for each real detection parameter
parameterTable: index to row of the reduced design matrix for each real detection parameter; dim(parameterTable) = c(uniquepar, np), where uniquepar is the number of unique combinations of paramater values (uniquepar < \(RnSKM\)) and np is the number of parameters in the detection model.
PIA: Parameter Index Array - index to row of parameterTable for a given session, animal, occasion and detector; dim(PIA) = c(R,n,S,K,M)
R: number of sessions

If models is empty then all components are NULL except for PIA which is an array of 1's (M set to 1).

Optionally (keep.dframe = TRUE) -

dframe: dataframe of design data, one column per covariate, one row for each c(R,n,S,K,M). For multi-session models n, S, and K refer to the maximum across sessions
validdim: list giving the valid dimensions (n, S, K, M) before padding

For make.lookup, a list with components

lookup: matrix of unique rows
index: indices in lookup of the original rows

For insertdim, a vector with length prod(dims) containing the values replicated according to dimx.

Arguments

capthist: capthist object
models: list of formulae for parameters of detection
timecov: optional dataframe of values of time (occasion-specific) covariate(s).
sessioncov: optional dataframe of values of session-specific covariate(s).
groups: optional vector of one or more variables with which to form groups. Each element should be the name of a factor variable in the covariates attribute of capthist.
hcov: character name of an individual (capthist) covariate for known class membership in h2 models
dframe: optional data frame of design data for detection parameters
naive: logical if TRUE then modelled detection probability is for a naive animal (not caught previously); if FALSE then detection probability is contingent on individual's history of detection
CL: logical; TRUE for model to be fitted by maximizing the conditional likelihood
keep.dframe: logical; if TRUE the dataframe of design data is included in the output
full.dframe: logical; if FALSE then padding rows are purged from output dframe (ignored if keep.dframe = FALSE)
ignoreusage: logical; if TRUE any usage attribute of traps(capthist) is ignored
contrasts: contrast specification as for model.matrix
...: other arguments passed to the R function model.matrix
tempmat: matrix for which row lookup required
x: vector of character, numeric or factor values
dimx: vector of notional dimensions for x to fill in target array
dims: vector of notional dimensions of target array

Details

These are internal secr functions that you are unlikely ever to use.

Each real parameter is notionally different for each unique combination of session, individual, occasion, detector and latent class, i.e., for \(R\) sessions, \(n\) individuals, \(S\) occasions and \(K\) detectors there are potentially \(R \times n \times S \times K\) different values. Actual models always predict a much reduced set of distinct values, and the number of rows in the design matrix is reduced correspondingly; a parameter index array allows these to retrieved for any combination of session, individual, occasion and detector.

The keep.dframe option is provided for the rare occasions that a user may want to check the data frame that is an intermediate step in computing each design matrix with model.matrix (i.e. the data argument of model.matrix).

... may be used to pass contrasts.arg to model.matrix.

Examples

Run this code

secr.design.MS (captdata, models = list(g0 = ~b))$designMatrices
secr.design.MS (captdata, models = list(g0 = ~b))$parameterTable

## peek at design data constructed for learned response model
head(captdata)
temp <- secr.design.MS (captdata, models = list(g0 = ~b),
    keep.dframe = TRUE)
a1 <- temp$dframe$animal == 1 &  temp$dframe$detector %in% 8:10
temp$dframe[a1,]

## ... and trap specific learned response model
temp <- secr.design.MS (captdata, models = list(g0 = ~bk),
    keep.dframe = TRUE)
a1 <- temp$dframe$animal == 1 &  temp$dframe$detector %in% 8:10
temp$dframe[a1,]

## place values 1:6 in different dimensions
insertdim(1:6, 1:2, c(2,3,6))
insertdim(1:6, 3, c(2,3,6))