Affiliations: [a]
School of Automation and Electrical Engineering, University of Science and Technology, Beijing, China
| [b] Key Laboratory of Advanced Control of Iron and Steel Process (Ministry of Education), Beijing, China
Correspondence:
[*]
Corresponding author. Wang Ling, School of Automation & Electrical Engineering, University of Science and Technology Beijing, Beijing 100083, China. Tel./Fax: +86 10 62315167; E-mail: lingwang@ustb.edu.cn.
Abstract: Although the association classification approach based on frequent patterns has been recently presented, the majority of the methods proposed so far do not deal with the quantitative data directly, and also do not consider the problem of exploring these rules to predict the future behavior of certain variables based on some other known variables. In light of these issues, a new algorithm based on quantitative association rules tree(CRQAR-tree) that synergizes association classification and rule-based TS fuzzy inference is developed to generate the rule tree structure and realize the classification and regression prediction. The classification and regression quantitative association rules are built on the improved Apriori algorithm which offered an efficient way for frequent itemsets learning. To manage the model complexity without sacrificing its predictive accuracy, CRQAR-tree can effectively match the rules to predict new samples that have little contribution over time. The proposed approach is applied to UCI benchmark datasets and a real application, the simulation results show that the performance of the CRQAR-tree is better than other methods, so it is a promising classification and regression algorithm.
Keywords: Quantitative association rule tree, classification, TS fuzzy inference, regression
