Stav dette: Discretizing environmental data for learning Bayesian-networkclassifiers