基于模式增长方式的高效用模式挖掘算法

高效用模式挖掘是数据挖掘领域的一个重要研究内容; 由于其计算过程包含对模式的内、外效用值的处理, 计算复杂度较大, 因此挖掘算法的主要研究热点问题就是提高算法的时间效率.针对此问题, 本文给出一个基于模式增长方式的高效用模式挖掘算法HUPM-FP, 该算法可以从全局树上挖掘高效用模式, 避免产生候选项集.实验中, 采用6个典型数据集进行实验, 并和目前效率较好的算法FHM (Faster high-utility itemset mining)做了对比, 实验结果表明本文给出的算法时空效率都有较大的提高, 特别是时间效率提高较大, 可以达到1个数量级以上.

An Algorithm for Mining High Utility Patterns Based on Pattern-growth

High utility pattern mining is an important research topic in data mining. Because of the additional inner/outer utility processing workload, its computational complexity increases, and the improvement of its temporal efficiency is vital. To address this issue, a new pattern-growth mining algorithm is proposed for high utility pattern mining named HUPM-FP. This algorithm can mine high utility patterns from a global tree without generating candidate itemsets. Six classical datasets were used in our experiments for comparing with the state-of-art algorithm faster high-utility itemset mining (FHM). The proposed HUPM-FP out-performed its counterpart significantly, especially for time efficiency, which was up to 1 order of magnitude faster.