Cluster Analysis is an indispensable part of data mining. It has a wider application now days. The paper comprises of rough set approaches wherein indiscernibility relation for similarity measures plays an important role in cluster analysis. Such type of relation gives rise to indiscernible classes. A similarity measure can represent the similarities between two objects. The object to be clustered is used in the decision on whether to put them into same cluster or disjoint cluster. In the proposed approach indiscernibility is used as a measure of similarity without any distance function for clustering the object. The concept of indiscernibility defines in a canonical rough set theory is relaxed. The proposed approach is carried in two phases. The first phase is considered to form all the identical groups together to form base clusters. In this case all the attribute values of the objects that belong to the same cluster are identical. While in the second stage, the strict notion of indiscernibility is relaxed and the classes are formed on the basis that objects are intra class similar or inter class dissimilar. The paper furnishes the example of mechanical engineering in which the working positions of machines are clustered with indiscernibility based relations.
Clustering Object, Indiscernibility Relation, Machine Learning Information, Rough Sets, Similarity Measures
