Creating control parameters for population size estimation and respective standard error and variance estimation.
Usage
controlPopVar(
alpha = 0.05,
bootType = c("parametric", "semiparametric", "nonparametric"),
B = 500,
confType = c("percentilic", "normal", "basic"),
keepbootStat = TRUE,
traceBootstrapSize = FALSE,
bootstrapVisualTrace = FALSE,
fittingMethod = c("optim", "IRLS"),
bootstrapFitcontrol = NULL,
sd = c("sqrtVar", "normalMVUE"),
covType = c("observedInform", "Fisher"),
cores = 1L
)Arguments
- alpha
significance level, 0.05 used by default.
- bootType
bootstrap type. Default is
"parametric", other possible values are:"semiparametric"and"nonparametric".- B
number of bootstrap samples to be performed (default 500).
- confType
type of confidence interval for bootstrap confidence interval,
"percentile"by default. Other possibilities:"studentized"and"basic".- keepbootStat
boolean value indicating whether to keep a vector of statistics produced by bootstrap.
- traceBootstrapSize
boolean value indicating whether to print size of bootstrapped sample after truncation for semi- and fully parametric bootstraps.
- bootstrapVisualTrace
boolean value indicating whether to plot bootstrap statistics in real time if
cores = 1ifcores > 1it instead indicates whether to make progress bar.- fittingMethod
method used for fitting models from bootstrap samples.
- bootstrapFitcontrol
control parameters for each regression works exactly like
controlMethodbut for fitting models from bootstrap samples.- sd
indicates how to compute standard deviation of population size estimator either as: \[\hat{\sigma}=\sqrt{\hat{\text{var}}(\hat{N})}\] for
sqrt(which is slightly biased if \(\hat{N}\) has a normal distribution) or fornormalMVUEas the unbiased minimal variance estimator for normal distribution: \[\hat{\sigma}=\sqrt{\hat{\text{var}}(\hat{N})} \frac{\Gamma\left(\frac{N_{obs}-1}{2}\right)}{\Gamma\left(\frac{N_{obs}}{2}\right)} \sqrt{\frac{N_{obs}}{2}}\] where the ration involving gamma functions is computed by log gamma function.- covType
type of covariance matrix for regression parameters by default observed information matrix.
- cores
for bootstrap only, number of processor cores to be used, any number greater than 1 activates code designed with
doParallel,foreachandparallelpackages. Note that for now using parallel computing makes tracing impossible sotraceBootstrapSizeandbootstrapVisualTraceparameters are ignored in this case.