挖掘支持度和兴趣度最优的数量关联规则

Mining Optimized Support and Interestingness Quantitative Association Rules

讨论了数量关联规则提取过程中的连续属性离散化方法和规则的有趣性问题 ,给出了数量关联规则的客观兴趣度的度量函数 ,提出用模板匹配方法挖掘用户感兴趣的规则 ,以解决数量关联规则有趣性的主观评测 ,研究了一种挖掘支持度和兴趣度最优的形如 (A∈ [v1 ,v2 ]∧ C1 ) C2 (其中 A为连续属性 ,C1 、C2 为类别属性 )的数量关联规则方法 ,并将该方法应用于股市行情分析 ,实验结果表明是非常有效的 .

This paper discusses the discreteness method for numeric attributes in the processing of mining quantitative association rules and the interestingness of association rules. An objective interestingness function of rule is given. For subjective interestingness measure, templates are used to differentiate the rules interested to the users or not. An algorithm for mining optimized support and interestingness quantitative association rules is presented. In the algorithm, we address the quantitative association rules such as (A∈[v 1,v 2]∧C 1)C 2, in which A is quantitative attribute, C1 and C2 are categorical attributes. The process of mining optimized support and interestingness quantitative association rules includes the several sub-processes, firstly partitioning for numeric attributes, secondly computing the maximum support ranges, finally selecting the most interesting rule from the maximum support rules. There are two examples in the paper, which applying the algorithm to analysis the buying and selling of stocks. The examples demonstrated the algorithm represented in this paper is practicable and effective.