Outlier Detection Using Rough Sets

Clustering-based methods for outlier detection are preferred in many contemporary applications due to the abundance of methods available for data clustering. However, the uncertainty regarding the cluster membership of an outlier object needs to be handled appropriately during the clustering process. Addressing this issue, this chapter delves on soft computing methodologies based on rough sets for clustering data involving outliers. In specific, the case of data comprising categorical attributes is looked at in detail for carrying out outlier detection through clustering by employing rough sets. Experimental observations on benchmark data sets indicate that soft computing techniques indeed produce promising results for outlier detection over their counterparts.