mi2.mar: Multiple-imputation-based (MI) abundance estimation under the...

In ecnuliuyang/CRAbun: Maximum likelihood abundance estimation from capture-recapture data under the Huggins-Alho model

Description

This function is used to calculate the MI estimate and Wald-type confidence interval of the abundance under the H-A model, as well as the estimates of parameters in the H-A model.

Usage

mi2.mar(d, K, x = NULL, y, level = 0.95,
  M = 100, seed = 321, beta.initial = NULL)

Arguments

d	a vector, the number of times of being captured.
K	a number, the number of capture occasions.
x	a vector, matrix or data.frame, stands for individual covariates without missingness.
y	a vector, matrix or data.frame, stands for individual covariates with missing values (NA).
level	a number, the nominal confidence level (default: 0.95).
M	a number, the number of imputated samples (default: 100).
seed	a number, seed used to produce repeatable results (default: 321).
beta.initial	a vector, with the same length as the parameters in H-A model. 0 means that the coefficient is fixed to zero, and non-zero values are seen as initial values when optimizing.

Details

Options d, K and y are necessary. Note that y must be univariate and the missing values.

Caution should be taken that beta.initial is NULL if all covariates are considered. When some coefficients of always observed covariates are zero, please specify the corresponding components of beta.initial as zero and others as none-zero values.

Value

A list containing the following components:

n.big	the MI estimate of the abundance.
n.big.se	Std. Error of n.big.
n.big.ci	the confidence interval of the abundance.
beta	the MI estimates of parameters in the H-A model.
beta.se	Std. Error of beta.

References

Lee, S.-M., W.-H. Hwang, and J. de Dieu Tapsoba (2016). Estimation in closed capture-recapture models when covariates are missing at random. Biometrics 72(4), 1294–1304.

Examples

## Analysis results of the prinia data set.

attach(prinia)

## Model 4
(mi4 <- mi2.mar( d = number.of.capture, K=17, x = cbind(1, fat.index, wing.index), y = tail.length,
                 beta.initial = c(-10,  0,  0, 0.1) ) )
## Model 5
(mi5 <- mi2.mar( d = number.of.capture, K=17, x = cbind(1, fat.index, wing.index), y = tail.length,
                 beta.initial = c(-10,  1,  0, 0.1) ) )
## Model 6
(mi6 <- mi2.mar( d = number.of.capture, K=17, x = cbind(1, fat.index, wing.index), y = tail.length,
                 beta.initial = c(-10,  0,  1, 0.1) ) )
## Model 7
(mi7 <- mi2.mar( d = number.of.capture, K=17, x = cbind(1, fat.index, wing.index), y = tail.length,
                 beta.initial = c(-10,  1,  0.5, 0.1) ) )

detach(prinia)

ecnuliuyang/CRAbun documentation built on April 13, 2020, 7:45 p.m.