Discovering Unordered Rule Sets for Mixed Variables Using an Ant-Miner Algorithm

Abstract

This work proposes a data mining algorithm called Unordered Rule Sets using a continuous Ant-Miner algorithm. The goal of this work is to extract classification rules from data. Swarm intelligence (SI) is a technique whereby rules may be discovered through the study of collective behavior in decentralized, self-organized systems, such as ants. The Ant-Miner algorithm, first proposed by Parpinelli and his colleagues (2002), applies an ant colony optimization (ACO) heuristic to the classification task of data mining to discover an ordered list of classification rules. Ant-Miner is a rule-induction algorithm that uses SI techniques to form rules. Ant-Miner uses a discretization process to deal with continuous attributes in the data. Discretization transforms numeric attributes into nominal attributes. Discretization may suffer from a loss of information, as the real relationship underlying individual values of a numeric attribute is unknown. The objective of this work is to apply ACO heuristic techniques to discover unordered rule sets for mixed variables in a data set. The proposed algorithm handles both nominal and continuous attributes using multimodal functions. It has the advantage of discovering more modular rules, i.e., rules that can be interpreted independently from other rules - unlike the rules in an ordered list, where the interpretation of a rule requires knowledge of the previous rules in the list. The results provide evidence that the accuracy of the Unordered Rule Set Continuous Ant-Miner algorithm is competitive with other Ant-Miner versions and generates simpler rule sets.