sicig {ig}R Documentation

Schwartz information criterium (SIC) for a sample of the IGTD

Description

The function sicig() gives the SIC value assuming an IGTD with parameters mu, lambda and a specific kernel.

Usage

sicig(x, kernel = "normal", nu.fixed = 2)

Arguments

x Vector of observations.
kernel Kernel of the pdf of the associated symmetrical distribution by means of which the IGTD is obtained. The kernels: "Laplace", "logistic", "normal" and "t" are available.
nu.fixed Additional parameter of the IGTD when the t kernel is used. This parameter corresponds to a shape parameter and it is also known as "degree of freedom". For default nu=1, in which case the Cauchy distribution is obtained. The Student-t distribution has always degrees of kurtosis greater than normal distribution. This aspect is transferred to the IGTD and produces robust parameter estimates for the IGTD.

Details

The SIC is a selection model criterion based on information loss. According to this criterion, it is possible to choice a hypothetic model that better describe the data set considering the smaller SIC value. The SIC is defined as SIC = -l(theta)/n+ p log(n)/(2n), where l(theta) is the log-likelihood function associated with the model, n is the sample size, and p is the number of involved parameters; for more details see Spieglhaiter, Best, Carlin and van der Linde (2002).

Author(s)

Víctor Leiva <victor.leiva@uv.cl>, Hugo Hernández <hugo.hernandez@msn.com>, and Antonio Sanhueza <asanhue@ufro.cl>.

References

Spieglhaiter, D. J., Best, N. G., Carlin, B. P., van der Linde, A. (2002). Bayesian measures of complexity and fit. Journal of the Royal Statistical Society Series B 64, 1-34.


[Package ig version 1.0 Index]