Temporal approach to association rule mining using T-tree and P-tree

Abstract

The real transactional databases often exhibit temporal characteristic and time varying behavior. Temporal association rule has thus become an active area of research. A calendar unit such as months and days, clock units such as hours and seconds and specialized units such us business days and academic years, play a major role in a wide range of information system applications. The calendar-based pattern has already been proposed by researchers to restrict the time-based associationships. This paper proposes a novel algorithm to find association rule on time dependent data using efficient T-tree and P-tree data structures. The algorithm elaborates the significant advantage in terms of time and memory while incorporating time dimension. Our approach of scanning based on time-intervals yields smaller dataset for a given valid interval thus reducing the processing time. This approach is implemented on a synthetic dataset and result shows that temporal TFP tree gives better performance over a TFP tree approach. © Springer-Verlag Berlin Heidelberg 2005.