时序波动关联规则概念格并行挖掘的一个新算法  

New Algorithm for Association Rule Mining in Trend Fluctuation Based on Concept Lattice

作　　者：吴大岳[1] 谢福鼎[1] 孙岩[1] 张永[1] 

机构地区：[1]辽宁师范大学计算机信息与技术学院,大连116029

出　　处：《计算机科学》2008年第10期158-161,共4页Computer Science

基　　金："973"项目(2004CB318000)资助;国家自然科学基金(10771092)资助

摘　　要：基于概念格理论,研究了对时间序列的波动情况进行周期关联规则的挖掘。首先对时间序列进行了反季节化预处理,然后提出了生成周期关联规则的新并行算法,在算法内部对生成的概念进行了剪枝处理,有效地提高了挖掘速度。随后用所给出的高精度模型对不满足移动平均法反季节化预处理条件的时间序列进行了反季节化计算。实验证明了该方法的有效性。This paper presented a method of cycle association rules mining for time-series fluctuations based on the concept lattice theory. The time-series were first deseasonalized and then a new parallel algorithm of cycle association rules mining was proposed. To increase the mining speed efficiently, some concepts obtained in the algorithm were pruned. Afterward,deseasonalizeation of time-series which do not satisfy the pretreatment conditions of deseasonalizeation in moving average method was computed by a high accuracy model given in this paper. The experiment results show the validity of this method.

关 键 词：概念格 时间序列 预测模型 周期关联规则 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] TP301.6[自动化与计算机技术—计算机科学与技术]