Dynamic information and library processing
Dynamic information and library processing
Concepts of the cover coefficient-based clustering methodology
SIGIR '85 Proceedings of the 8th annual international ACM SIGIR conference on Research and development in information retrieval
Toward experimental evaluation of subsystem classification recovery techniques
WCRE '95 Proceedings of the Second Working Conference on Reverse Engineering
Approaches for measuring the stability of clustering methods
ACM SIGIR Forum
Improving clustering stability with combinatorial MRFs
Proceedings of the 15th ACM SIGKDD international conference on Knowledge discovery and data mining
Hi-index | 0.00 |
Among the significant factors in assessing the suitability of a clustering technique to a given application is its stability; that is, how sensitive the algorithm is to perturbations in the input data. A number of techniques that appear to be suitable for measuring the stability of clustering have been published in the literature. The details about each of these measures, such as a description of the steps involved in their computation and an identification of precisely what they measure, are presented. These measures are considered in the context of analysing the stability characteristics of clustering techniques and are compared using a framework developed for this purpose. The question of generalizing some of these measures is addressed and the measures are also analyzed to identify conditions under which they can be reduced to one another.