Analyzing data clusters: A rough sets approach to extract cluster-defining symbolic rules

Abstract

We present a strategy, together with its computational implementation, to intelligently analyze the internal structure of inductivelyderived data clusters in terms of symbolic cluster-defining rules. We present a symbolic rule extraction workbench that leverages rough sets theory to inductively extract CNF form symbolic rules from un-annotated continuous-valued data-vectors. Our workbench purports a hybrid rule extraction methodology, incorporating a sequence of methods to achieve data clustering, data discretization and eventually symbolic rule discovery via rough sets approximation. The featured symbolic rule extraction workbench will be tested and analyzed using biomedical datasets.