1

Does anyone know if there is a simple way of generating predictions from an ASreml-R model for cases where the response variable is missing but all values for fixed effects and levels of random effect factors are known. The generic "fitted" function returns predictions for cases with observed responses only.

The "predict" function I would use to generate predicted responses over a new data set from a model fitted in in e.g. lmer does not seem to work with asreml.

I am not sure that the predict.asreml function will do quite what I want but perhaps I am misunderstanding how to use it.

As ASReml is not open source I haven't provided a reproducible example.

Thanks, Patrick

Kevin Wright
  • 2,397
  • 22
  • 29
PS1
  • 11
  • 2

1 Answers1

1

I had similar problem. Here is my quick fix, try:

coef(model.asr)$fixed

and

coef(model.asr)$random

to extract the variable and add them.

Linxiao Bai
  • 21
  • 2