Abstract—Time series pattern recognition in motif discovery has emerged as one of the prominent primitives for data series mining and knowledge discovery. In the context of mining for scalable financial time series patterns, the repeated price patterns are not only unknown in advance, but they are also highly scalable in size, which require a more responsive elastic distance measure than lock-step distance measure. Nonetheless, the state-of-the-art motif discovery techniques are still operating on motif pair of equal length with lock-step Euclidean distance metric. Computational complexity in time series search is the bottleneck that not only refrains brute-force search with different length, but also deprives the implementation of more complex elastic distance measure. Hence, this study introduced a novel double-cycled Value Based Data Representation approach with inherent time transformation element from the data representation process. It aims to produce elastic measure comparable result by only using Euclidean distance metric, bypassing the computational complexity presence in elastic distance measure.
Index Terms—Data representation, motif discovery, time series pattern recognition, time series data mining.
K. H. Sim, K. Y. Sim, and V. Raman are with the Swinburne University of Technology Sarawak, Jalan Simpang Tiga, 93350 Kuching, Sarawak, Malaysia (e-mail: email@example.com, firstname.lastname@example.org, email@example.com).
Cite: Kwan-Hua Sim, Kwan-Yong Sim, and Valliappan Raman, "Scalable Pattern Recognition in Financial Time Series Data with Double-Cycled Value Based Data Representation," International Journal of Machine Learning and Computing vol. 10, no. 3, pp. 416-422, 2020.Copyright © 2020 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).