On Learning Decision Rules From Flow Graphs
Chien-Chung Chan, Shusaku Tsumoto · 2007
The use of flow graphs to represent information flow distribution from data tables for intelligent data analysis was first proposed by Pawlak. This paper studies the representation of flow graphs by multiset decision tables. This representation is minimal. Inspired by the flow graphs, a new rule learning algorithm based on this representation is presented with examples. Two sets of rules are learned from certain examples and examples in the boundary set. Rules are characterized by Bayesian factors introduced by Pawlak.