Bug in the transformation function

[qu-cheng]

obj$marginals.hyperpar[[1]] at Line 75 should be obj$marginals.hyperpar[[1 + spat.cor.param]]? Otherwise, it will refer to the gaussian observation precision for INDPMCAR and PMCAR.

https://github.com/becarioprecario/INLAMSM/blob/706d8340e973407b7e0eb18d582f59ef9b7b0a65/R/transform.R#L70-75

[qu-cheng]

I realized that the original code works well for Poisson regression, but not linear regression

[becarioprecario]

I realized that the original code works well for Poisson regression, but not linear regression

Yes. The function is tailored to work with a model with no other hyperparmeters. Perhaps we should explain this in the documentation and mention to be cautious. The function could be used as a template for a more general model as well.