A generalized rand-index method for consensus clustering of separate partitions of the same data base

One of the recent trends in industry-based cluster analysis, especially in marketing, is the development of different partitions (e.g., needs-based, p sycho-graphics, brand choice, etc.) of the same set of individuals. Such in dividualized clusterings are often designed to serve different objectives. Frequently, however, one would also like to amalgamate the separate cluster ings into a single partition-one that parsimoniously captures commonalities among the contributory partitions. In short, the problem entails finding a consensus partition of T clusters, based on J distinct, contributory parti tions (or, equivalently, J polytomous attributes). We describe a new model/ algorithm for implementing this objective. The method's objective function incorporates a modified Rand measure, both in initial cluster selection and in subsequent refinement of the starting partition. The method is applied to both synthetic and real data. The performance of the proposed model is c ompared to latent class analysis of the same data set.