Melanoma is a prevalent skin cancer in Australia, with 14320 new cases estimated to be diagnosed in 2018. Survival times are markedly different from one individual to the next, however there appears to be three classes of survival outcome. One interest of many researchers in the field is to investigate whether prognosis of Melanoma patients can be predicted from patient gene expression data. Our goal is to be able to build a classifier that not only accurately predicts prognosis class, but also provides meaningful and interpretable conclusions regarding the features chosen. We construct a hybrid model that seamlessly integrates a multi-class diagonal discriminant analysis model and variable selection components. Our variable selection component naturally simplifies as a function of likelihood ratio statistics allowing natural comparisons with traditional hypothesis testing methods. We compare our method with several competing approaches with both simulation results (based on simulation conditions proposed by Tibshirani and Witten (2011)), and other biological datasets.