Mining fuzzy rules from quantitative data based on the Variable Precision Rough Set and Rough Fuzzy Set Theories
Hongkang Lin - · Journal of Convergence Information Technology · 2011
In this paper, a different approach to extract the threshold value β of Variable Precision Rough Set (VPRS) applied to continuous information systems is presented. This study combines the Fuzzy Set and Rough Fuzzy Set (RFS) theories to determine the β value of VPRS. The β value was determined by the Fuzzy C-means and relevant Fuzzy theories, for the reason that errors of system classification could occur in the fuzzy-clustering phase prior to information classification. In practice, the attributes of the continuous information system were first fuzzed and classified by Fuzzy C-means method, then the maximum of the membership function value was chosen as the cluster forming basis. Finally, the β value of approximate classification was obtained under the VPRS by Implication Relations and Rough Fuzzy Set (RFS) theory. In addition, the influences of Implication Relations of various definitions on the β-lower approximation of VPRS were discussed.