Category Selection for Multinomial Data
Rebecca M. Baker, Frank P. A. Coolen · 2009
A new method is presented for selecting a single cate-gory or the smallest subset of categories, based on ob-servations from a multinomial data set, where the se-lection criterion is a minimally required lower proba-bility that (at least) a specific number of future obser-vations will belong to that category or subset of cat-egories. The inferences about the future observations are made using an extension of Coolen and Augustin’s nonparametric predictive inference (NPI) model to a situation with multiple future observations.