APDREcorr: APDREcorr Correlations between Activity Patterns using APD...

In AnimalAPD: Compare Activity Patterns with Activity Probability Density (APD)

Description

Calculates estimated relationships between activity probability density (APD) of the focal and contingent(s) using Bayesian GLMMs with 'STAN' using brm, with the option to automatically select the statistical distribution that best fits the dataset (weibull, frechet, gamma, lognormal, inverse gaussian) by loo. The function automatically ensures that MCMC chains reach convergence and that the specified minimum effective sample size from the posterior distribution is achieved.

Package: AnimalAPD Version: 1.0.0 Date: 2020-11-10

Usage

APDREcorr(
  focal,
  cont1,
  cont2 = NULL,
  cont3 = NULL,
  cont4 = NULL,
  RE1,
  RE2 = NULL,
  weibullGLMM = TRUE,
  frechetGLMM = TRUE,
  gammaGLMM = TRUE,
  lognormalGLMM = FALSE,
  invgaussianGLMM = TRUE,
  cores = 1,
  iter = 5000,
  minESS = 1000,
  burnin = iter/2,
  thin = 1,
  adapt_delta = 0.95,
  center = "pi",
  adjust = 1,
  Reloo = TRUE,
  plothist = TRUE,
  ploteffects = TRUE,
  histcol = "cyan4",
  effectcol = "cyan4",
  linecol = "red"
)

Arguments

focal	Vector of observations in radians of one species/group/individual/etc. for which predictions on another will be made.
cont1	Vector of observations in radians, or output from generalized circular mixture model of activity curves from GCMM, of a species/group/individual/etc. from which predictions will be made
cont2	Optional vector of observations in radians, or output from generalized circular mixture model of activity curves from GCMM, of additional species/group/individual/etc. from which predictions will be made
cont3	Optional vector of observations in radians, or output from generalized circular mixture model of activity curves from GCMM, of additional species/group/individual/etc. from which predictions will be made
cont4	Optional vector of observations in radians, or output from generalized circular mixture model of activity curves from GCMM, of additional species/group/individual/etc. from which predictions will be made
RE1	Vector identifying a random intercept for observations of the focal to control for hierarchical data (e.g. camera trap IDs)
RE2	Optional vector identifying levels of a second random effect, for data with additional hierarchical levels (e.g. study sites, sampling periods, data collection seasons); default is NULL
weibullGLMM	Specifies whether to run a weibull GLMM, using the brms package; default is TRUE for all and results from the best-fitting model are returned
frechetGLMM	Specifies whether to run a frechet GLMM, using the brms package; default is TRUE for all and results from the best-fitting model are returned
gammaGLMM	Specifies whether to run a Gamma GLMM, using the brms package; default is TRUE for all and results from the best-fitting model are returned
lognormalGLMM	Specifies whether to run a lognormal GLMM, using the brms package; default is TRUE for all and results from the best-fitting model are returned
invgaussianGLMM	Specifies whether to run a inverse.gaussian GLMM, using the brms package; default is TRUE for all and results from the best-fitting model are returned
cores	Number of cores to use when running MCMC chains in parallel; default=1
iter	Number of MCMC iteractions per chain; burnin is iter/2; default=5000
minESS	Desired minimum effective sample size; default=1000
burnin	Number of MCMC iterations to be discarded as the burn-in; default=iter/2
thin	Thinning rate for saving MCMC draws; default=1
adapt_delta	Value to use for adapt_delta with brms; default=0.95; see also brms
center	Value to use as center of graph; default=pi
adjust	Smoothing of predicted line; recommended to use default value for observed values and higher value for estimations from circular models
Reloo	Whether to use reloo when running leave-one-out cross-validation of models (loo); see also brms and loo
plothist	Whether to plot histograms of samples from the posterior distribution for the correlation parameters; default=TRUE
ploteffects	Whether to plot predicted effects; default=TRUE
histcol	Colour for histogram bars
effectcol	Colour for predicted effect plot 95% HDI
linecol	Colour for histogram lines for the 95% HDI and 0

Value

Prints results of best-fitting model and posterior samples and/or predicted effects of parameter estimates if plothist=TRUE and ploteffects=TRUE, and returns object of class APD with list of analysis results and information.

data List of data used in analysis

model Object of class brmsfit containing results and information for best-fitting model.

distribution Character vector of statistical distribution of best-fitting model

allmodels List of output for all tested models; object of class brmsfit

Author(s)

Liz AD Campbell

See Also

GCMM brm loo

Examples

data(wolfexample)
data(boarexample)
APDREcorr(focal=wolfexample$Radians,cont1=boarexample$Radians,
    RE1=wolfexample$SamplingPeriod)

AnimalAPD documentation built on March 18, 2021, 5:07 p.m.