Abstract: A method of rule induction based on rough set theory,which is capable of learning both deterministic and non-deterministic classification rules,is discussed.The method makes the classification rules more general and reliable.An algorithm is put forward.The basic idea is to allow users to specify three parameters:strength,consistency and coverage,then the classification rules that satisfy the requirements are induced.At the end of the paper,the rules which meet the specified criteria are derived by applying the method to an example.The method is robust in dealing with noisy data and in analyzing large databases.

Key words: rough sets, variable precision rough set, information system, induction of probabilistic rules