PREPPY: An Approach for Rule Induction in the DRSA Context

Complex decision-making environments demand multicriteria methods aligned with the decision maker’s preferences. In this article, we present PREPPY, a DRSA rule induction algorithm which, in an incremental and top-down manner, specializes condition sets guided by information gain, precision, and recall, while exploring permutations of promising criteria. The method integrates quantitative and qualitative attributes with flexible orderings, avoiding artificial discretizations and mitigating combinatorial explosion through stopping criteria. After induction, redundant rules are pruned and conflicts are resolved by prioritizing higher precision and, subsequently, coverage, thus preserving conciseness and interpretability. Preliminary tests suggest stable models with good generalization ability, although computational cost increases on large datasets, motivating optimizations, further validations, and evaluation under variable consistency.