An Approach to Rough Set Decomposition of Incomplete Information Systems

Incomplete information systems exist broadly in practical data analysis, and approaches to complete the incomplete information system through various completion methods in the preprocessing stage are normal in data mining. These methods may result in distortion of original data and knowledge, and can even render the original data mining system un-minable. To sidestep these shortcomings inherent in traditional methods, this paper proposed a decomposition approach for incomplete information system. This approach does not require completion of the system beforehand, but it choose a template based on rough set evaluation functions and then extract complete subsets from the incomplete information system hierarchically with the template. Then an intermediate variable is constructed based on rough sets theories and is used to decompose incomplete information system to simplify rule sets. The obtained rule sets are then applied layer by layer to reasoning and decision analysis. A concrete example of this approach applied to diagnostic data of vibration faults for steamer-generator units is given in this paper, to validate the effectiveness of this algorithm in processing incomplete information systems.