sbaic: Akaike's Information Criterion

In scaleboot: Approximately Unbiased P-Values via Multiscale Bootstrap

Description

Extract or modify the AIC values for models.

Usage

sbaic(x,...)
## S3 method for class 'scaleboot'
sbaic(x,k,...)
## S3 method for class 'scalebootv'
sbaic(x,...)

sbaic(x) <- value
## S3 replacement method for class 'scaleboot'
sbaic(x) <- value
## S3 replacement method for class 'scalebootv'
sbaic(x) <- value

Arguments

x	an object used to select a method.
k	numeric, the penalty per parameter to be used.
value	numeric vector of AIC values for models.
...	further arguments passed to and from other methods.

Details

sbaic can be used to modify the aic components for models in x as shown in the examples below.

Value

For an object of class "scaleboot", sbaic returns a numeric vector of AIC values for models. If k is missing, then the aic components in the fi vector of x are returned. If k is specified, rss-k*df is calculated for each model. For the usual AIC, k=2. For the BIC (Schwarz's Bayesian information criterion), k=log(sum(x$nb)).

Author(s)

Hidetoshi Shimodaira

References

Sakamoto, Y., Ishiguro, M., and Kitagawa G. (1986). Akaike Information Criterion Statistics. D. Reidel Publishing Company.

See Also

sbfit.

Examples

data(mam15)
a <- mam15.relltest[["t4"]] # an object of class "scaleboot"
sbaic(a) # print AIC for models
sbaic(a,k=log(sum(a$nb))) # print BIC for models
sbaic(a) <- sbaic(a,k=log(sum(a$nb))) # set BIC
sbaic(a) # print BIC for models

scaleboot documentation built on Dec. 4, 2019, 5:07 p.m.