A Rough Set Model Based on Probabilistic Similarity Measure for Incomplete Decision Tables

Rough set models in incomplete decision tables have been discussed so far. Numerous approaches to deal with missing values in incomplete information systems have been proposed. In this paper, assuming that the domain of attribute values is defined, we apply the probability of values appearing in data tables in order to measure the self-information of similarity. This is defined as the uncertainty of similarity. Based on that, set approximation is defined by giving a threshold. Finally, the merit of this similarity is clarified and compared with other approaches.