基于多最小支持度的增量式关联规则挖掘算法研究

Research on Incremental Updating Algorithm of Association Rules Mining Based on Multiple Minimum Supports

在传统的Apriori关联规则挖掘算法分析基础上,针对目前多最小支持度和增量式关联规则挖掘的局限性,提出基于多最小支持度的增量式关联规则挖掘算法。该算法适用于事务出现频率一致及不一致的情况,利用多最小支持度能挖掘出更有意义的结果;同时,该算法还能实现事务数据不断增加时的数据挖掘,提高了挖掘的效率。应用电力客户信用数据库进行实验的结果表明,改进算法能有效挖掘出稀有项,分析出潜在的信用风险客户,对电力客户信用评价具有辅助决策作用。

By analyzing the classical Apriori algorithm of association rules mining, and in view of the limitations of multiple minimum supports and incremental association rules mining, the incremental updating algorithm of association rules mining based on multiple minimum supports is put forward. The proposed algorithm is suitable for different frequency of services and ever-increasing amount of data, thus improving the efficiency of mining. Experiments were made with the crediting database of power customers, and the result shows that the improved algorithm can effectively find out the rare rules and support decision-making on crediting evaluation of power customers through analysis of potential delinquent customers.