Fis a MARSS model to data from each ESUs treating each population as a subpopulation. The structure of the variance-covariance matrix, the U matrix, the Z matrix, and the R matrix can be specified. If you want to fit a specific model, then pass in model as a list as per a MARSS model. The populations in the ESU with < min.years of data points are not used in the fitting and no states are estimated for those.
trend_fits( datalist, outputfile, wild = TRUE, model = NULL, logit.fw = TRUE, min.years = 5 )
| datalist | The list output by data_detup() |
|---|---|
| outputfile | The name of the RData file to save the results to. |
| wild | wild=TRUE means to do the fit on fracwild*total versus on the total spawners. |
| model | If null, a set of models is fit. Otherwise pass in a model specified as a list in MARSS format. |
| logit.fw | If TRUE fit to logit of fracwild instead of the raw percentages. |
| min.years | Only populations with at least min.years will be used in the fitting. |
A list with three items:
A list with the fits for each ESUs included.
If there are multiple models fit, then the AIC will be returned.
If there are multiple models fit, then the best model is returned.
If model=NULL then a set of all possible models is fit. This takes awhile but will allow one to use AIC to compare the model set. wild=TRUE means to do the fit on fracwild*total versus on the total spawners. logit.fw says whether to fit to logit of fracwild or to the percentages.
This function produces a states estimate and a fracwild fit;
Eli Holmes, NOAA, Seattle, USA. eli(dot)holmes(at)noaa(dot)gov