Efficient mining frequent closed resource patterns in resource effectiveness data: The MFPattern approach

Abstract

As the occurrence of failure of electronic resources is sudden, real-time record analysis on the effectiveness of all resources in the system can discover abnormal resources earlier and start using backup resources or restructure resources in time, thus managing abnormal situations and finally realizing health management of the system. This paper proposed an algorithm for mining frequent closed resource patterns from data effectiveness matrix with the method of column extension: MFPattern, which uses effective pruning strategies to guarantee the mining of all frequent closed patterns without producing candidate item-sets. Different from the traditional frequent closed pattern, MFPattern algorithm can mine resource combination patterns with all resources very effective during work, those with simultaneous failure of resources and combination patterns in which some resources are very effective while some other resources have failure. The experimental result shows that this algorithm has a higher mining efficiency than existing mining methods of frequent closed pattern. © 2014 Springer-Verlag.