Fit an edge model to data

Usage

bison_model(
  formula,
  data,
  model_type = c("binary", "count"),
  directed = FALSE,
  partial_pooling = FALSE,
  zero_inflated = FALSE,
  priors = NULL,
  refresh = 0,
  mc_cores = 4,
  iter_sampling = 1000,
  iter_warmup = 1000,
  priors_only = FALSE
)

Arguments

formula: Formula specifying social events and sampling effort on the LHS and edge weights, fixed, and random effects on the RHS.
data: Aggregated or disaggregated dataframe of dyadic observations.
model_type: "binary" or "count", specifying the type of edge weight model to use.
directed: TRUE or FALSE specifying whether the network is directed or not.
partial_pooling: Whether to pool edge weights so that information is shared between edges.
zero_inflated: Whether to use a zero-inflated model to model excess zeroes.
priors: List of priors in the format supplied by get_default_priors().
refresh: Frequency of messages printed while running the sampler.
mc_cores: Number of cores to use when running the sampler.
iter_sampling: Number of iterations to use for posterior samples.
iter_warmup: Number of iterations to use for warmup (will not be used for samples).
priors_only: Whether to use priors as posteriors or to allow the posteriors to be updated by data.

Value

An S3 edge model object containing edge samples and processed data.

Details

Fits a BISoN edge weight model to a user-provided dataframe. The function supports either aggregated (at the dyad-level) or disaggregated (at the observation-level) dataframes. Node names or IDs need to be formatted as factors with the same levels.

The type of edge model and the interpretation of edge weights used depends on model_type, and will change the interpretation of the edge weights.