基于距离的关联规则挖掘算法研究

Improvement of the Distance-based Association Rule Algorithms

提出了基于距离的关联规则算法的几点改进:在聚类部分,改用CADD算法对全部属性聚类,使得聚类结果更好,并且减少了规则的判定条件;在关联度参数D0的设置问题上,提出了用投影簇半径值作为其参考值的思想,以减少设置的盲目性.实验结果表明,改进的算法能更有效地挖掘基于距离的关联规则.

Some improvements about the algorithm of distance-based association are proposed. On the part of the clustering, all properties by CADD algorithm are clustered to make the results better,and to reduce the rules to determine requirements; On the issuc of the correlation parameter Do＇s setting,the idea of cluster radius as its reference value is proposed in order to reduce the set of blindness. The experimental results show that the improved algorithm can be moi＇e effctive in the distance-based association rules mining.