面向频繁封闭渐进项集的挖掘算法

主流数据挖掘算法不能有效解决大规模数值数据集挖掘问题。提出了一种应用于大规模数值数据集的线性时间封闭项集挖掘改进(Improved Linear time Closed Itemsets Minner, ILCM)算法。ILCM算法使用能够提取属性共同变化量的渐进模式挖掘方法,借鉴LCM算法的前缀保留闭合扩展思想,通过深度优先搜索输出频繁封闭渐进项集结果。实验证明,相比传统挖掘算法,ILCM能够显著提高算法运行效率和降低内存空间占用,并且能够有效处理如DNA微阵列等实际大型数值数据集挖掘。

Research on Mining Algorithm towards Closed Frequent Gradual Item sets

The current data mining algorithms cannot be used to solve the large scale numerical dataset. A novel algorithm called Improved Linear time Closed Item sets Minner (ILCM) is proposed to be applied for the scalable numerical datasets. Based on ppc (prefix preserving closure)-extension principle of the standard LCM, ILCM uses gradual pattern mining method with the capability of extracting the covariation of attributes, and then outputs the result of closed frequent gradual item sets according to depth first searching method. The experimental demonstrates that, compared to the traditional mining algorithm, ILCM can improve the running efficiency and reduce memory space occupation significantly, and ILCM is proven to apply in the practical large scale numerical dataset such as DNA micro-array data.