Mining Time Series Data with Two Dimensional Fuzzy Pattern Rules
Haifeng Xia, Bing Chen, Jiawei Fan, Zhi Li, Dan Gao
Available Online August 2015.
- https://doi.org/10.2991/msam-15.2015.78How to use a DOI?
- fuzzy pattern rule; referential vector distance; K-means cluster; J-measure;stage characteristics
- Based on association rules and fuzzy pattern theory, this paper presents a data mining model to extract patterns using two-dimensional fuzzy pattern rules, focusing on the discretization of the time series of dependent and independent variables. The referential vector distance is established as a similarity function, and K-means clustering method is used to acquire the basic trend features. The silhouette coefficient is used to test and analyze the clustering result. Due to the potential delay of the influence of the independent variable on the dependent variable, we determine the optimal lag phase with the sort rule of J-measure based on the principle of maximum entropy. The fuzzy pattern rules of the influence of the independent variable on the dependent variable are determined and interpreted according to the trend features and the fuzzy pattern definitions. The obtained character sets are adjusted and iterated several times based on the definitions of the stage characteristics to locate and interpret the inherent periodic characteristics of the couple of time series. The reliability of the proposed model is tested with an empirical analysis of the effect of the price of corn on that of the pork.More importantly, the analysis unit of this model can be as accurate as the time of a day, suggesting its potential advantages in analyzing the short-term characteristics of time series.
- Open Access
- This is an open access article distributed under the CC BY-NC license.
Cite this article
TY - CONF AU - Haifeng Xia AU - Bing Chen AU - Jiawei Fan AU - Zhi Li AU - Dan Gao PY - 2015/08 DA - 2015/08 TI - Mining Time Series Data with Two Dimensional Fuzzy Pattern Rules PB - Atlantis Press SP - 342 EP - 349 SN - 1951-6851 UR - https://doi.org/10.2991/msam-15.2015.78 DO - https://doi.org/10.2991/msam-15.2015.78 ID - Xia2015/08 ER -