Influence diagnostics in a class of extreme-value BS regression model
This work introduces a local influence graphical analysis on a class of extreme-value Birnbaum- Saunders regression models. They are an extension based on the skew-normal distribution of the usual Birnbaum-Saunders (BS) regression and are useful for analyzing data of extreme events. The proposal uses conformal normal curvature (CNC) of the log-likelihood function to diagnose the influence of observations on the postulated model. Eigenvalues and eigenvectors associated with the conformal normal curvature may help determine which direction of the perturbation is influential via a graphical representation. The numerical illustration includes a brief simulation study. Finally, we have also analyzed a data set to illustrate the use of the methodology.