Advanced approach of sliding window based erasable pattern mining with list structure of industrial fields.

As one of the various data mining areas, erasable pattern mining is a method for solving financial problems that can be caused in various industrial fields. Since the concept of erasable pattern mining was proposed, various relevant techniques have been developed. Dealing with dynamic stream data is an important issue in different data mining areas including erasable pattern mining. In order to deal with partial data with the latest information, a sliding window-based approach was proposed. However, the method still has performance limits in many cases because of its inefficient data structures and mining techniques. Motivated by this problem, we propose a new efficient erasable pattern mining algorithm based on the sliding window that can guarantee robustness and reliability and deal with the latest stream data more effectively. In addition, the newly proposed list data structure and mining techniques allow the proposed method to extract erasable patterns in a more efficient way. Results of empirical experiments using well-known benchmark datasets guarantee that the proposed algorithm outperforms previous state-of-the-art approaches in various aspects.