Towards Efficient Discovery of Partial Periodic Patterns in Columnar Temporal Databases

Penugonda Ravikumar, Bathala Venus Vikranth Raj, Palla Likhitha, Rage Uday Kiran, Yutaka Watanobe, Sadanori Ito, Koji Zettsu, Masashi Toyoda

Abstract: Finding partial periodic patterns in temporal databases is a challenging problem of great importance in many real-world applications. Most previous studies focused on finding these patterns in row temporal databases. To the best of our knowledge, there exists no study that aims to find partial periodic patterns in columnar temporal databases. One cannot ignore the importance of the knowledge that exists in very large columnar temporal databases. It is because real-world big data is widely stored in columnar temporal databases. With this motivation, this paper proposes an efficient algorithm, Partial Periodic Pattern-Equivalence Class Transformation (3P-ECLAT), to find desired patterns in a columnar temporal database. Experimental results on synthetic and real-world databases demonstrate that 3P-ECLAT is not only memory and runtime efficient but also highly scalable. Finally, we present the usefulness of 3P-ECLAT with a case study on air pollution analytics.