一种改进的增量关联规则算法

An Improved Incremental Updating Algorithm for Association Rules

针对增量式更新关联规则算法FUP会产生大量候选项集和多次扫描数据库的问题,提出改进算法NFUP。该算法通过新旧数据库频繁项集间的关系得出所有频繁项集,尽可能利用已有的挖掘结果来生成较少的候选项目集并较少次数地扫描数据库。仿真实验表明,NFUP算法的执行时间比FUP算法减少了不少。

Fast Update（FUP） algorithm always produces a lot of candidate itemsets and scans database many times, so a new improved Fast Updating （NFUP） algorithm is presented to solve such two bottleneck problems of FUP. This algorithm produces fewer candidate itemsets and fewer number of times scanning database, because it obtains all frequent item sets through frequent itemsets＂ relations of new and old database and uses the existing results as far as possible. Simulation shows that the execution time of NFUP algorithm reduces a lot compared with FUP algorithm in the case of the same database.