S3 methods

[Tess-LaCoil]

@fontikar I am preparing for re-submission to JStatSoft, and one of their comments was that the hmde_object class does not have default methods. This is the class assigned to output of the hmde_model function on line 27 of hmde_model.R:

class(output) <- "hmde_object"

The hmde_object returned is typically used as input for hmde_assign_data, but the latter function does not check if the passed model_template is of the correct type. My question is, do we need to retain this class assignment to preserve good structure in the package? I can write the required functions easily enough.

If we do want to preserve it, I will change the class name to hmde_model_template, have the assign data function check that it is passed one, and write S3 functions for summary, plot, and print.

@dfalster At present there is no requirement to have S3 methods for the fit object, because it's a stanfit object rather than a custom one. However, it may be useful to do so, so I'm going to create S3 function templates for a hmde_fit object, and change the output class to it.

[Tess-LaCoil]

Current thinking for what to return for each:
hmde_model_template class:

• Summary and print both return the list structure, with type info
• Plot returns a plot of the chosen DE with parameters set to the prior means.

hmde_fit class:

• Summary and print return a table of top-level parameter values (name, mean, median, 95% CI), some model fit info eg. R-hat distribution.
• Plot returns a grid of top-level parameter chain plots, histogram of R-hats. There are 7 of these for the Canham multispecies (6 sp-level parameters, global error parameter)