A Real Time Hybrid Pattern Matching Scheme for Stock Time Series

Zhang, Z., Jiang, J., Liu, X., Lau, R., Wang, H. and Zhang, R.

Pattern matching in stock time series is an active research area in data mining. We propose a new real-time hybrid pattern-matching algorithm in this paper. The algorithm is based on the Spearman’s rank correlation, rule sets and sliding window. The concept of sliding windows enables patterns matching to be performed only based on subsequence of stock data which are freshly received. Therefore the proposed algorithm can be applied in real-time application and processing time can be reduced. Spearman’s rank correlation coefficient is used to classify the preferred patterns effectively and efficiently first and use the rule sets to provide further ability for describing the query patterns so that is more effective, sensitive and constrainable in distinguishing individual patterns. Encouraging experiment is reported from the tests that the proposed scheme outperforms the other methods both effectively and efficiently, especially in differentiating the special preferred stock patterns or even distorted patterns.

Cite as: Zhang, Z., Jiang, J., Liu, X., Lau, R., Wang, H. and Zhang, R. (2010). A Real Time Hybrid Pattern Matching Scheme for Stock Time Series. In Proc. 21st Australasian Database Conference (ADC 2010) Brisbane, Australia. CRPIT, 104. Shen H.T. and Bouguettaya, A. Eds., ACS. 161-170

(from crpit.com) (local if available)