基于分段函数的多支持度关联规则挖掘算法

Association rules mining algorithm with multiple supports based on piecewise function

在多支持度关联规则挖掘算法中,针对最小支持度的选取问题,提出一种基于分段函数的多支持度关联规则挖掘算法。在多支持度算法中挖掘频繁集的时候,最小支持度由项集最小项支持度的最小值、最大值和给定的参考值所决定,这样避免了采用最小值作为最小支持度算法的时间复杂度高和存在无效规则的问题,以及采用最大值致使剪枝程度过大而造成规则遗漏的问题。通过实验结果表明了该算法的有效性。

Aimed at how to select the minimum support in the association rules mining algorithms with multiple supports, a new mining algorithm with multiple supports is proposed, which is based on piecewise function. The minimum support is determined by the minimum, maximum support of the itemset and the reference value that is given in the mining process of finding frequent itemset. The problems are avoided that are the high algorithm time complexity and existing invalid rules with using the minimum support of itemset as the minimum support, and the issues are avoided that are eliding useful rules because of the excessive pruning with using the maximum of itemset as the minimum support. Finally, the experimental results show the new algorithm is efficient.