hmmpars.paramtetric.boot: Parameteric bootstrap availability HMM.

In david-borchers/hmltm: Line transect estimation with hidden Markov availability models

Parameteric bootstrap availability HMM.

Description

Bootstraps hidden Markov model (HMM) data from estimated parameters and their variance-covariance matrix (one of each for each animal).

Usage

hmmpars.paramtetric.boot(availhmm, animals, seed = NULL, B, printprog = TRUE)

Arguments

availhmm	list with availability HMM parameters, as for the hmm.pars parameter of est.hmltm. The list must contain these elements: Pi: a 3D matrix of Markov chain transition probability matrices, with element [,,i] containing the 2D matrix (Pi) for animal i (state 1=unavailable, state 2=available). pm: a matrix with element [,i] containing the state-dependent Bernoulli “success” parameters (p) for each state (first state=unavailable, 2nd=available) for animal i. delta: a matrix with element [,i] containing the stationary disribution of the Markov chain (first state=unavailable, 2nd=available). vcv: a 3D array with element [,,i] containing the estimated variance- covariance matrix of the following (in this order): logit(Pi[1,1]), logit(Pi[2,1]),logit(p[1]),logit(p[2]).
animals	a vector of integers indicating which of the elements of adat are to be resampled.
seed	random number seed (integer).
B	number of bootstraps to perform (integer).
printprog	if TRUE prints progress through animals as it resamples.

Details

Resamples HMM paramters assuming asymptotic normality of parameters. Constructs a new hmm.pars object from the fitted HMM parameters.

Does the above B times, each time reformtting the hmm.pars object as a vector using vectorize.hmmpars and then entering this as the next row in a matrix of dimension BxT, where T=3+length(animals)*(nstate^2+nstate*2) and nstate is the number of states in the HMM.

Value

A matrix of dimension BxT, where T is as described above.

david-borchers/hmltm documentation built on Oct. 29, 2023, 9:07 p.m.