一种改进的基于距离的孤立点挖掘算法的研究

摘要

对经典的基于距离的孤立点挖掘算法进行了改进,引入"关键属性",即减少了挖掘的数据量,从而提高了孤立点挖掘的效率,并且改进常用距离度量,用改进的加权曼式距离进行计算,降低了数据分布不均给检测结果带来的影响,同时改进后的挖掘算法在不影响挖掘结果的情况下取消了对参数设置的要求,给出了孤立点孤立的程度.%By introducing " key attribute" , the reduction of the data to be dug so as to raise the outliers digging efficiency, improvement is made on algorithm for distance - based outlier digging. By using the improved weighted Man - type calculation of distance, detection results tend to be more accurate, and the improved algorithm requires no settings to the parameter without affecting the digging effect. The extent, to which an outlier is isolated, is also provided.