ABSTRACT
Suppose that we are given a square asymmetric data matrix O ¼ (ojk) for a set of objects where ojk represents the measure of relationship between a pair of objects.
In order to explore some structure inherent in the data or derive some information
from them, we can utilize procedures that are provided without referring to the
meaning of the measure, in other words, whether the measure represents simi-
larity or dissimilarity. In this chapter we describe those procedures independent
of whatever the measure means. At the observation of a phenomenon, the diag-
onal entries may or may not be defined for the matrix O ¼ (ojk), or some of the diagonal entries may be missing. In such cases, we fill zero values in those diag-
onal elements and will treat O with all entries defined in what follows.
