Efficient mining regularly frequent patterns in transactional databases.

Abstract

Finding interesting patterns plays an important role in several data mining applications, such as market basket analysis, medical data analysis, and others. The occurrence frequency of patterns has been regarded as an important criterion for measuring interestingness of a pattern in several applications. However, temporal regularity of patterns can be considered as another important measure for some applications. In this paper, we propose an efficient approach for miming regularly frequent patterns. As for temporal regularity measure, we use variance of interval time between pattern occurrences. To find regularly frequent patterns, we utilize pattern-growth approach according to user given min_support and max_variance threshold. Extensive performance study shows that our approach is time and memory efficient in finding regularly frequent patterns.