An Ordering Algorithm for Analysis of Data Arrays

Abstract

This paper presents an algorithm for reordering data arrays. The object is to find a one-dimensional ordering of the rows and columns of an array, given no prior information other than the array itself. The algorithm is an iterative one, and, in finding what it considers to be the most satisfactory ordering of an array, it also enables one to determine immediately the most significant underlying variable. The algorithm has been applied to arrays representing data from a variety of fields; examples are presented of political and archaeological data that have been analyzed, and it is shown that meaningful relations can be found directly through use of the algorithm.