摘要： 介绍关联规则挖掘中的经典算法——Apriori算法的关键思想。针对传统Apriori算法效率上的不足，提出一种改进的Apriori算 法——Ec-Apriori算法。该算法采用划分的方法，按照频繁1-项集支持度对数据库进行划分，在各自独立的子数据库上运用等价类的方法生成候选集，优化连接操作，同时利用位对象操作简化支持度的计算，较好地提高了算法效率。实验结果表明，改进后的算法具有较好的有效性。

关键词: 关联规则Apriori算法, 数据库划分, 等价类, 位对象

Abstract: This paper introduces the principle of the Apriori algorithm which is the classical algorithm of association rules mining, and proposes an improved Apriori algorithm――Equivalence class-Apriori(Ec-Apriori) algorithm which aims at the disadvantage of Apriori algorithm. It adopts a partition method and partitions the database into some independent databases according to the support of frequent 1-itemsets. It generates the candidate itemsets by adopting equivalence class in separate databases and optimizes the join operation, and simplifies the account of support by bit object, so it is more efficient. Experimental result shows that the Ec-Apriori algorithm outperforms Apriori algorithm, and gets a good practicality.

Key words: association rules, Apriori algorithm, database partition, equivalence class, bit object

中图分类号: 

• TP301.6