一种基于密度的离群点检测方法

A Density-based Outlier Detection Method

基于局部密度的差异来发现离群点的检测方法很难处理离群点聚集在一起的情况,提出一种基于密度的离群点检测方法,该方法先采用DBSCAN聚类算法检测出全局离群点,然后借鉴局部离群因子的评估策略来确定大类簇边界区域内的"错聚"样本点,进而从"错聚"样本点的邻居点中依据距离和局部密度识别出其他局部离群点。实验结果表明该方法具有一定的可行性和有效性。

Outlier detection methods based on the difference between the local density of sample points have difficulty dealing with the case that outliers get together.The proposed method was first applied in the DBSCAN algorithm for global outlier detection,and then the boundary sample points clustered into the wrong cluster were identified by the local outlier factor.At last,other local outlier points within the neighborhood of the boundary points were recognized by measuring the distance and local density.Experimental results show that the proposed method is feasible and effective.