Damped window based high average utility pattern mining over data streams

摘要: Abstract Data mining methods have been required in both commercial and non-commercial areas. In such circumstances, pattern techniques can be used to find meaningful information. Utility (UPM) is more suitable for evaluating the usefulness of patterns. The method introduced this paper employs high average utility (HAUPM) approach, which one UPM approaches discovers interesting patterns items relations among another by using a novel measure. Meanwhile, past research on algorithms mainly focus tasks processing static database as batch operations. Most continuous, unbounded stream data constantly produced from heart beat sensors should treated differently with respect importance because up-to-date may higher influence than old data. Therefore, our approach also adopts concept damped window model gain useful environments. Various experiments are performed real datasets order demonstrate that designed not only provides important, recent information but requires less computational resources execution time, memory usage, scalability significant test.