An Optimal Rule Set Generation Algorithm for Uncertain Data

Nowadays, mining of knowledge from large volumes of datasets with uncertainties is a challenging issue. Rough Set Theory (RST) is the most promising mathematical approach for dealing with uncertainties. This paper proposes an RST-based optimal rule set generation (ORSG) algorithm for generating the optimal set of rules from uncertain data. At first, the ORSG approach applies the concepts of RST for identifying inconsistencies and then from the preprocessed consistent data, the RST-based Improved Quick Reduct Algorithm is used to find the most promising feature set. Finally, from the obtained prominent features, the Reduct-based Rule Generation algorithm generates the optimal set of rules. The performance of the ORSG approach is 10-fold cross-validated by conducting experiments on the UCI machine learning repository’s Thyroid disease dataset, and the results revealed the effectiveness of the ORSG approach.