Extension of rough set methodology for probabilistic decision analysis from preferential multiple attribute decision tables

Based on substitution of the indiscernibility relation by a dominance relation and a confident threshold value set, an extension of the rough sets model is proposed, which can deal with possible inconsistencies in preferential multiple attribute decision tables and obtain preferential probabilistic decision rules made up of preference actions. Finally, a kind of algorithm for mining preferential probabilistic rules is presented. Results demonstrate that the minimal sets of preferential probabilistic rules derived from the model have a smaller number of stronger rules and a smaller number of the use conditions.