An efficient data structure for decision rules discovery

Abstract

The increasing amount of information available is encouraging the search for efficient techniques to improve the data mining methods, especially those which consume great computational resources. We present a novel structure, called EES, which helps the data mining algorithms which generate decision rules to reduce the aforementioned cost. Given that decision rules establish conditions for database attributes, EES stores the information in such a way that the search can be carried out by attributes instead of by examples. EES could be useful for any method which generates decision rules. Moreover, it is of particular interest when the search for the solution involves a great many hypothetical solutions. Thus, this structure is designed for speeding up the rule-evaluation process in methods based on Evolutionary Algorithms. The traditional structure, based on vectors of examples (in which the database is stored) is evaluated and compared with EES, including the costs for a stratified set of cases. Finally, the experimental results demonstrate the quality of our proposal, reducing the computational cost by approximately 50%.