Fit a TVVARSS model to multivariate time series data

tvvarss is the primary function for fitting TVVARSS models data.

tvvarss(
  y,
  de_mean = TRUE,
  topo = NULL,
  dynamicB = TRUE,
  family = "gaussian",
  x0 = NULL,
  shared_q = NULL,
  shared_r = NULL,
  process = NULL,
  mcmc_iter = 1000,
  mcmc_warmup = 500,
  mcmc_thin = 1,
  mcmc_chain = 3,
  ...
)

Arguments

y	The data (array, with dimensions = site, year, species)
de_mean	Whether or not to de_mean the process model; defaults to TRUE. For example, \(X_{t+1} = B_{t} (X_{t} - pred[X_{t}])\) versus \(X_{t+1} = B_{t} X_{t}\).
topo	Optional list matrix describing the presumed topology of the community. Pairwise interactions are specified as density-dependent ("dd"), top-down ("td"), bottom-up ("bu"), competitive/facilitative ("cf"), or absent ("zero").
dynamicB	Logical indicator of whether to fit a dynamic B matrix that varies through time (or a static B matrix that does not); defaults to TRUE.
family	Statistical distribution for the observation model, defaults to "gaussian". But can be any of "gaussian", "binomial", "poisson", "gamma", "lognormal"
x0	The location matrix (mean) of priors on initial states; defaults to centered on observed data.
shared_q	Optional matrix (number of species x number of sites) with integers indicating which process variance parameters are shared; defaults to unique process variances for each species that are shared across sites.
shared_r	Optional matrix (number of species x number of sites) with integers indicating which observation variance parameters are shared; defaults to unique observation variances for each species that are shared across sites.
process	Vector that optionally maps sites to states. Defaults to each site as its own state
mcmc_iter	Number of MCMC iterations, defaults to 1000
mcmc_warmup	Warmup / burn in phase, defaults to 500
mcmc_thin	MCMC thin, defaults to 1
mcmc_chain	MCMC chains, defaults to 3
...	Extra arguments to pass to sampling

Value

an object of class 'stanfit'