Learn R Programming

⚠️There's a newer version (3.0) of this package.Take me there.

hierarchicalDS (version 2.9)

Functions For Performing Hierarchical Analysis of Distance Sampling Data

Description

Functions for performing hierarchical analysis of distance sampling data, with ability to use an areal spatial ICAR model on top of user supplied covariates to get at variation in abundance intensity. The detection model can be specified as a function of observer and individual covariates, where a parametric model is supposed for the population level distribution of covariate values. The model uses data augmentation and a reversible jump MCMC algorithm to sample animals that were never observed. Also included is the ability to include point independence (increasing correlation multiple observer's observations as a function of distance, with independence assumed for distance=0 or first distance bin), as well as the ability to model species misclassification rates using a multinomial logit formulation on data from double observers. New in version 2.1 is the ability to include zero inflation, but this is only recommended for cases where sample sizes and spatial coverage of the survey are high.

Copy Link

Version

Install

install.packages('hierarchicalDS')

Monthly Downloads

18

Version

2.9

License

Unlimited

Maintainer

Paul B Conn

Last Published

November 26th, 2014

Functions in hierarchicalDS (2.9)

function to calculate the joint pdf for a sample of values from one of a number of pdfs

function to export posterior summaries from an mcmc object to a table

Produce an adjacency matrix for a square grid

log_lambda_log_likelihood

compute the likelihood for nu parameters

plot 'observed' versus predicted values for abundance of each species at each transect

function to stack data (going from three dimensional array to a two dimensional array including only "existing" animals

function to sample from a specified probability density function

function to simulate double observer spatial distance sampling data subject to possible zero inflation and species misidentification

generate initial values for MCMC chain if not already specified by user

convert.HDS.to.mcmc

function to convert HierarchicalDS MCMC list vector (used in estimation) into an mcmc object (cf. coda package)

get_confusion_array

Fill confusion array - one confusion matrix for each individual (DEPRECATED)

function to plot a map of abundance. this was developed for spatio-temporal models in mind

MCMC output from running example in Hierarchical DS

Produce an RW2 Adjacency matrix for a rectangular grid for use with areal spatial models. This formulation uses cofficients inspired by a thin plate spline, as described in Rue & Held, section 3.4.2 Here I'm outputting an adjacency matrix of 'neighbor weights' which makes Q construction for regular latices easy to do when not trying to make inference about all cells (i.e., one can just eliminate rows and columns associated with cells one isn't interested in and set Q=-Adj+Diag(sum(Adj))

function to produce a design matrix given a dataset and user-specified formula object

log_lambda_gradient

compute the first derivative of log_lambda likelihood component for Langevin-Hastings

switch_sample_prior

function to sample from hyperpriors of a specified probability density function; note that initial values for sigma of lognormal random effects are fixed to a small value (0.05) to prevent numerical errors

Mrds probit detection and related functions

Produce an RW1 adjacency matrix for a rectangular grid for use with areal spatial models (queens move)

MCMC output from running example in Hierarchical DS

generate_inits_misID

generate initial values for misID model if not already specified by user

estimate optimal 'a' parameter for linex loss function

calculate parameter estimates and confidence intervals for various loss functions

get_confusion_mat

Fill a list with confusion matrices for each record

stack_data_misID

function to stack data for midID updates (going from four dimensional array to a two dimensional array including observed groups

function to calculate posterior predictive loss given the output object from hierarchicalDS

SIMULATE AN ICAR PROCESS

Produce an adjacency matrix for a vector

Function for MCMC analysis

hierarchical_DS

Primary function for hierarchical, areal analysis of distance sampling data. This function pre-processes data and calls other functions to perform the analysis, and is the only function the user needs to call themselves.