Cluster quality assessment by the modified RenyiClip X algorithm
Dalia Baziuk · 2010
This paper presents the modified RenyiCLIPx clustering algorithm and shows that with this algorithm more accurate groupings can be obtained, it gives higher accuracy rates as well. The combination of Renyi entropy based clustering and CLIP3, CLIP4 rule derivation algorithms is used to discover clusters and create rules to explain them. The algorithm itself contains refinements that are used while improving the goodness of obtained clusters and rules. Results on researching the influence of data discretization and so called negative examples data set to the rule complexity and cluster quality are presented as well.