A new inter cluster validation method for unsupervised clustering techniques

In this paper, a new cluster validation technique called Inter Cluster Validation (ICV) is presented. The proposed technique is aimed to measure the inter cluster similarity and inter cluster dissimilarity over the each individual cluster with other clusters in the resulting cluster of the unsupervised clustering techniques. The proposed ICV technique consists of two measures which are Inter Cluster Similarity (ICS) and Inter Cluster Dissimilarity (ICD). The first measure (ICS), is aimed to measure the inter cluster similarity over the each individual cluster with other cluster in resulting cluster and, it also calculate the overall resulting cluster similarity. The second measure (ICD), is evaluate the inter cluster dissimilarity over the each individual with other clusters in the resulting cluster. The proposed ICV technique is test over the resulting cluster of the four unsupervised clustering techniques are K-Means, CURE, OAC and LIAC. Experimental results show that the ICV technique is simple and better suitable for evaluating the inter cluster similarity and inter cluster dissimilarity over the each individual cluster belongs to the resulting cluster of the unsupervised clustering techniques such as K-Means, CURE, OAC and LIAC.