Compute AICc for one or several fitted model objects for which a log-likelihood value can be obtained.
AICc(object, ..., k = 2)If just one object is provided, a numeric value with the corresponding AICc.
If multiple objects are provided, a data.frame with rows corresponding
to the objects and columns representing the number of parameters estimated
(df) and the AICc.
A fitted model object from splm(), spautor(), spglm(), or spgautor()
where estmethod is "ml" or "reml".
Optionally more fitted model objects.
The penalty parameter, taken to be 2. Currently not allowed to differ from 2 (needed for generic consistency).
When comparing models fit by maximum or restricted maximum
likelihood, the smaller the AICc, the better the fit. The AICc contains
a correction to AIC for small sample sizes. The theory of
AICc requires that the log-likelihood has been maximized, and hence,
no AICc methods exist for models where estmethod is not
"ml" or "reml". Additionally, AICc comparisons between "ml"
and "reml" models are meaningless -- comparisons should only be made
within a set of models estimated using "ml" or a set of models estimated
using "reml". AICc comparisons for "reml" must
use the same fixed effects. To vary the covariance parameters and
fixed effects simultaneously, use "ml".
Hoeting et al. (2006) study AIC and AICc in a spatial context, using the AIC definition
\(-2loglik + 2(estparams)\) and the AICc definition as
\(-2loglik + 2n(estparams) / (n - estparams - 1)\), where \(n\) is the sample size
and \(estparams\) is the number of estimated parameters. For "ml", \(estparams\) is
the number of estimated covariance parameters plus the number of estimated
fixed effects. For "reml", \(estparams\) is the number of estimated covariance
parameters.
spmod <- splm(z ~ water + tarp,
data = caribou,
spcov_type = "exponential", xcoord = x, ycoord = y
)
AICc(spmod)
AIC(spmod)
BIC(spmod)
Run the code above in your browser using DataLab