wbonat
commented
3 years ago Hi
Concerning predictions it depends on your model. If you have simple models for independent observations the predictions are easily obtained using the linear predictor. However, I do not have a generic method for the entire class, that's the reason I do not provide a prediction function.
In relation to simulations, the mcglm is based only on second moments, basically it means that we do not have a multivariate distribution available to simulate from it. Thus, I have used the NORTA (Normal to Anything) to simulate from mcglm models. It is a reasonable solution, but has limitations. I have used the NORTARA package in R to do it, but the specification of the NORTA algorithm is case specific, that's why the mcglm does not provide a generic function to simulate data from mcglm models.
Since, simulations are not implemented in the mcglm package I do not provide any documentation on it, but you can have a look at the NORTARA documentation's package.
Best wishes.
Dear authors, thanks for building the package and it has a great many applications in real-world modeling. This might be a stupid question but I couldn't find answers online - Is there a predict function to generate prediction on new data with mcglm models? Furthermore, if I want to run simulations on new data using mcglm models, are there any examples on how to do so? Thanks!