Utilization of attribute clustering methods for scalable computation of reducts from high-dimensional data
Andrzej Janusz, Dominik Ślȩzak · QUT ePrints (Queensland University of Technology) · 2012
We investigate methods for attribute clustering and their possible applications to a task of computation of decision reducts from information systems. We focus on high-dimensional data sets, for which the problem of selecting attributes that constitute a reduct can be extremely computationally intensive. We apply an attribute clustering method to facilitate construction of reducts from microarray data. Our experiments confirm that by proper grouping of similar, in some sense replaceable attributes it is possible to significantly decrease a computation time, as well as increase a quality of resulting reducts (i.e. decrease their average size).