APDRE: Activity Probability Density Controlling for Random Effects
In AnimalAPD: Compare Activity Patterns with Activity Probability Density (APD)

Description Usage Arguments Value Author(s) See Also Examples

View source: R/AnimalAPD.R

Calculation of animal activity probability density controlling for nested data with random intercepts using Bayesian GLMMs with 'STAN' and brm. The function can automatically select the statistical distribution that is most appropriate for the dataset (weibull, frechet, gamma, lognormal, inverse gaussian) using loo and automatically ensures that MCMC chains converge and that a specified minimum effective sample size from the posterior distribution is achieved. An APD activity curve plot is provided.

Package: AnimalAPD Version: 1.0.0 Date: 2020-11-08

APDRE(
  focal,
  contingent,
  RE1,
  RE2 = NULL,
  weibullGLMM = TRUE,
  frechetGLMM = TRUE,
  gammaGLMM = TRUE,
  lognormalGLMM = FALSE,
  invgaussianGLMM = TRUE,
  cores = 1,
  iter = 5000,
  burnin = iter/2,
  center = "pi",
  Reloo = TRUE,
  adapt_delta = 0.95,
  adjust = 1,
  minESS = 1000,
  col = "deeppink3",
  histcol = "deeppink",
  linecol = "black",
  xlimCV = NULL,
  min = TRUE,
  max = TRUE,
  points = TRUE,
  mean = TRUE,
  HDI = TRUE,
  rawmean = FALSE,
  ...
)

`focal`	Vector of observations in radians of one species/group/individual/etc. for which predictions on another will be made
`contingent`	Vector of observations in radians or output from generalized circular mixture model of activity curves from `link[activityGCMM]{GCMM}` of a 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; default=5000
`burnin`	Number of MCMC iteractions discarded as burnin; default=iter/2
`center`	Value to use as center of graph; default=pi
`Reloo`	Whether to use reloo when running leave-one-out cross-validation of models (loo)
`adapt_delta`	Value to use for adapt_delta with brms; default=0.95
`adjust`	Smoothing of predicted line; recommended to use default value for observed values and higher value for estimations from circular models
`minESS`	Desired minimum effective sample size; default=1000
`col`	Specifies colour of points for the focal in the graph
`histcol`	Specifies colour of the histogram plot of the posterior distribution
`linecol`	Specifies colour of HDI line in histogram plot of the posterior distribution
`xlimCV`	A vector of two values indicating the x axis limits for the histCV graph
`min`	Whether to include minimum APD on the graph; default=TRUE; default=TRUE
`max`	Whether to include maximum APD on the graph; default=TRUE; default=TRUE
`points`	Whether to include datapoints for observations of the focal on the graph; default=TRUE
`mean`	Whether to include the estimated mean APD from the GLMM on the graph; default=TRUE
`HDI`	Whether to include the estimated 95% highest density interval of mean APD from the GLMM on the graph; default=TRUE
`rawmean`	Whether to include the raw mean, not correcting for random effects, on the graph; default=FALSE
`...`	Additional parameters

Prints graph of activity curve and APD estimates from best-fitting GLMM and prints summary of analysis. Returns object of class APD is returned, containing a list of analysis results and details:

data List of data used in analysis

output Matrix with summary output from selected model

distribution Name of distribution of selected model

model An object of class brmsfit containing output from the selected model, including the posterior samples and other information. See brm

CVposterior Numeric vector of posterior samples for the calculated family-specific population coefficient of variation (CV)

allmodels List of objects of class brmsfit containing output from all models from the analysis.

rawvalues Numeric vector of the raw, uncorrected APD values

rawsummary List of summary stats of raw APD values

Liz AD Campbell

GCMM brm loo

data(wolfexample)
data(boarexample)
 APDRE(focal=wolfexample$Radians, contingent=boarexample$Radians, 
    RE1=wolfexample$SamplingPeriod, weibullGLMM=TRUE, frechetGLMM=FALSE,
    gammaGLMM=FALSE, lognormalGLMM=FALSE, invgaussianGLMM=FALSE,
    min=TRUE, max=TRUE, points=TRUE, mean=TRUE, HDI=TRUE, rawmean=FALSE)