分组多支持度关联规则研究

Mining association rules with grouping multiple minimum supports

关联规则是数据挖掘的重要任务之一,传统关联规则算法只有一个最小支持度,假设项出现的频率大致相同,而在实际中并非如此,由此产生了多支持度关联规则问题。该问题针对每个项给定不同的支持度,而在实际应用中项可以划分成若干个组,每组有一个支持度。由此提出了分组多支持度关联规则问题,针对该问题给出了基于多支持度性质对项进行分组的方法。该方法可以降低2-项候选集的数目。在此基础上,进一步给出了相应的多支持度关联规则发现算法,并通过实验证明了算法的有效性。

Association rule mining is an important model in data mining, since only one minimum support is used in the traditional association rules mining model, it implicitly assumed that all items in the data had similar frequencies. This is seldom the case in real life applications, which is the origin of the multiple minimum supports association rules mining. However multiple minimum supports association rules mining supposed all item has different support, this is unusual in true-life, different items will belong to several groups, each group has different minimum support. This is root of grouping multiple minimum supports association rule mining. According to this, an algorithm for mining association rules with grouping multiple minimum supports is discussed. The property of multiple minimum supports is analyzed. According to the property, a method of grouping items is proposed, which reduce the number of the candidate 2-itemsets. Based on the grouping method, an algorithm is proposed, and experiment show the algorithm quiet efficient.