DIC: Deviance information criterion for a 'graf' model

In GRaF: Species distribution modelling using latent Gaussian random fields

Description

Calculates the deviance information criterion (DIC) and effective parameters for a graf model.

Usage

DIC(object)

Arguments

object

A graf object.

Details

Information criteria can be used to compare the goodness of fit of two models to the same dataset whilst accounting for model complexity. Commonly used methods such as AIC are difficult to implement for models such a GRaF where the number of parameters in the model is not obvious. Spiegelhalter et al. (2002) proposed the effective number of parameters for these models and the associated DIC for model comparison. Here we calculate DIC by numerical integration. If weights are included in the model fitting they are taken into account when calculating the DIC. DIC should therefore not be used to compare between models with different observation weights.

Value

A vector containing:

DIC	The deviance information criterion.
pD	The estimated number of effective parameters in the model.

References

Spiegelhalter, D.J.. Best, N.G., Carlin, B.P., van der Linde, A. (2002). Bayesian measures of model complexity and fit. Journal of the Royal Statistical Society, Series B, 64 (4): 583-639.

See Also

graf

GRaF documentation built on May 29, 2017, 9:50 a.m.