一种基于规则的离群挖掘算法

离群数据挖掘是指从大量数据中挖掘明显偏离、不满足一般行为模式的数据。现有的离群数据挖掘算法大多对密集的交易数据库缺乏有效的处理,文中提出了一种高效的基于规则的离群挖掘算法。该算法使用了多层最大离群支持度及最小离群兴趣度,计算1-离群条件集的幂集,并在数据结构中存储了交易标识符链表,使得扫描数据库的次数仅为一次,从而提高了挖掘的速度、效率且使得结果更具有决策意义。文中使用此算法对某一商场的部分销售数据库进行了实验,结果表明该算法能有效、迅速地发现密集数据库中的离群数据。

An Algorithm for Outliers Mining Based on Rule

Outlier mining refers to the mining of the data with obvious departure,and with no general behaviors patterns from large amounts of data.Most existing algorithms don't have good performance when dealing with data-intensive transaction databases.This paper presents a highly efficient rule-based outlier mining algorithm.This algorithm uses multi-layer maximum outlier support and minimum outlier interest,calculates the power of 1-outlier set,and uses linked list to store the transaction identifier in data structure.All these make the time of scanning database only once.So the mining speed and efficiency are enhanced,the result are more useful to decision-making.In this paper,an experiment using this algorithm was carried out on part of a shopping center's scales database.The result presents that this algorithm can mine data-intensive transaction database more effectively and rapidly.