Testing for revealing of data structure based on the hybrid approach

Abstract

In this paper testing for revealing data structure based on a hybrid approach has been presented. The hybrid approach used during the testing suggests defining a pre-clustering hypothesis, defining a pre-clustering statistic and assuming the homogeneity of the data under pre-defined hypothesis, applying the same clustering procedure for a data set of interest, and comparing results obtained under the pre-clustering statistic with the results from the data set of interest. The pros and cons of the hybrid approach have been also considered.