cAIC4 (version 1.0)

getWeights: Optimize weights for model averaging.

Description

Function to determine optimal weights for model averaging based on a proposal by Zhang et al. ( 2014) to derive a weight choice criterion based on the conditional Akaike Information Criterion as proposed by Greven and Kneib (2010). The underlying optimization is a customized version of the Augmented Lagrangian Method.

Usage

getWeights(models)

Arguments

models

An list object containing all considered candidate models fitted by lmer of the lme4-package or of class lme.

Value

An object containing a vector of optimized weights, value of the minimized target function and the duration of the optimization process.

WARNINGS

No weight-determination is currently possible for models called via gamm4.

References

Greven, S. and Kneib T. (2010) On the behaviour of marginal and conditional AIC in linear mixed models. Biometrika 97(4), 773-789.

Zhang, X., Zou, G., & Liang, H. (2014). Model averaging and weight choice in linear mixed-effects models. Biometrika, 101(1), 205-218.

Nocedal, J., & Wright, S. (2006). Numerical optimization. Springer Science & Business Media.

See Also

lme4-package, lmer, getME

Examples

# NOT RUN {
data(Orthodont, package = "nlme")
models <- list(
    model1 <- lmer(formula = distance ~ age + Sex + (1 | Subject) + age:Sex,
               data = Orthodont),
    model2 <- lmer(formula = distance ~ age + Sex + (1 | Subject),
               data = Orthodont),
    model3 <- lmer(formula = distance ~ age + (1 | Subject),
                 data = Orthodont),
    model4 <- lmer(formula = distance ~ Sex + (1 | Subject),
                data = Orthodont))

foo <- getWeights(models = models)
foo


# }