Rule induction for incomplete information systems

We proposed a modified rule generation algorithm (MRG) to generate a minimal set of rule reducts and proposed a generalized rule generation algorithm (MRGI) to generate a minimal set of rule directly from the original incomplete information system. Based on MRGI, with each rule reduct represents a unique decision rule. We developed a rule generation and rule induction prototype (RGRIPI) to extract certain rules directly from the incomplete information system. RGRIPI can automatically generate a minimal set of decision rules directly from an incomplete data set. We build a probability function combining the plausibility and probability of missing values to compute the possible rules for incomplete information systems.