GOFSN: Goodness-of-fit tests for the family of skew-normal models

GOFSN is a package that implements a method for checking if a skew-normal model fits the observed dataset, when all parameters are unknown. While location and scale parameters are estimated by moment estimators, the shape parameter is integrated with respect to the prior predictive distribution, as proposed in (BOX, 1980). A default and proper prior on skewness parameter is used to obtain the prior predictive distribution, as proposed in (CABRAS, CASTELLANOS, 2008). Goodness-of-fit tests, here proposed, depend only on sample size and exhibit full agreement between nominal and actual size. This package implements EDF statistics Kolmogorov-Smirnov(D), Cram\'er-von Mises(W2) and proposes some simple algorithms (SimulD,SimulW2) to approximate their respective marginal predictive distributions. It also has functions (ks.sn,W2.sn) that calculate the p-value on observed data.

Downloads: