fit: Fitting function to be called by user

In ericward-noaa/phenomix: Fit Density Curves to Peak Timing Data that Varies over Time

Fitting function to be called by user

Description

This function creates a list of parameters, sets up TMB object and attempts to do fitting / estimation

Usage

fit(
  data_list,
  silent = FALSE,
  inits = NULL,
  control = list(eval.max = 2000, iter.max = 1000, rel.tol = 1e-10),
  limits = NULL,
  fit_model = TRUE
)

Arguments

data_list	A list of data, as output from create_data
silent	Boolean passed to TMB::MakeADFun, whether to be verbose or not (defaults to FALSE)
inits	Optional named list of parameters for starting values, defaults to NULL
control	Optional control list for stats::nlminb. For arguments see ?nlminb. Defaults to eval.max=2000, iter.max=1000, rel.tol=1e-10. For final model runs, the rel.tol should be even smaller
limits	Whether to include limits for stats::nlminb. Can be a list of (lower, upper), or TRUE to use suggested hardcoded limits. Defaults to NULL, where no limits used.
fit_model	Whether to fit the model. If not, returns a list including the data, parameters, and initial values. Defaults to TRUE

Examples

data(fishdist)

# example of fitting fixed effects, no trends, no random effects
set.seed(1)
datalist <- create_data(fishdist[which(fishdist$year > 1970), ],
  asymmetric_model = FALSE,
  est_mu_re = FALSE, est_sigma_re = FALSE
)
fit <- fit(datalist)
#
# # example of model with random effects in means only, and symmetric distribution
# set.seed(1)
# datalist <- create_data(fishdist[which(fishdist$year > 1970),], asymmetric_model = FALSE,
#                      est_sigma_re = FALSE)
# fit <- fit(datalist)
# # example of model with random effects in variances
# set.seed(1)
# datalist <- create_data(fishdist[which(fishdist$year > 1970),], asymmetric_model = TRUE,
#                          est_mu_re = TRUE)
# fit <- fit(datalist)
#
# # example of model with poisson response
# set.seed(1)
# datalist <- create_data(fishdist[which(fishdist$year > 1970),], asymmetric_model = FALSE,
#                         est_sigma_trend=FALSE, est_mu_trend=FALSE, est_mu_re = TRUE,
#                         family="poisson")
# fit <- fit(datalist)

ericward-noaa/phenomix documentation built on May 6, 2024, 10:20 a.m.