Skip to contents

Find the AIC and BIC for a set of models fitted using asreml

Source: R/icREML.R
ic_reml_asr.Rd

The function is a wrapper for 'icREML' function described in Verbyla (2019).

Usage

ic_reml_asr(fm, scale = 1)

Arguments

fm

A list of asreml fitted model objects

scale

A scalar to scale the variance matrix of the estimated fixed effects (to ensure numerical stability of a log-determinant). Default value is 1.

Value

A data frame. The data frame has the following components

  • model : the names of the models

  • loglik : the full log-likelihood for each model

  • p : the number of fixed effects parameters for each model

  • q : the number of (non-zero) variance parameters for each model.

  • b : the number of variance parameters that are fixed or on the boundary. These parameters are not counted in the AIC or BIC.

  • AIC : the AIC for each model

  • BIC : the BIC for each model

  • logdet : the log-determinant used in adjusting the residual log-likelihood for each model

References

Verbyla, A. P. (2019). A note on model selection using information criteria for general linear models estimated using REML. Australian & New Zealand Journal of Statistics, 61(1), 39-50.

Author

Ari Verbyla (averbyla at avdataanalytics.com.au)