一种改进的Eclat算法

An Improved Eclat Algorithm

首先,深入分析了频繁模式挖掘算法Eclat和Eclat+,在大数据集上挖掘长模式时,Eclat+的性能不及Eclat。基于此,提出一种改进的Eclat算法,新算法充分利用了垂直数据表示和交叉计数的高效优势,直接在垂直数据表示的数据集上通过广度优先搜索和交叉计数产生频繁模式。实验结果表明,在挖掘长模式时,改进的Eclat算法的运行速度较Eclat、Eclat+均有明显的提高。

First, the two association rule algorithms Eclat and Eclat＋ are compared. Eclat＋ is not as good as Eclat when mining long patterns in large dataset. Then, it proposes a new improved association rule algorithm based on Eclat. The new algorithm is implemented by vertical data layout, breadth first search, and intersection. It makes use of the efficiency of vertical data layout and intersection. The new algorithm against Eclat and Eclat＋ is experimentally compared, making significant progress in runtime on our test database when mining long patterns in large dataset.