Methods of forming clusters for objects

Abstract

Great variety can be seen in the methods of forming clusters for objects. Most methods have been developed separately so that authors do not often build on the theories of others. Hence this chapter will not be very systematically structured.

The first techniques to be discussed require data measured on interval level; they take their point of departure (or can do so) from a matrix of D ²-coefficients. After this attention is turned to three hierarchical clustering schemes and to methods using concepts derived from information theory. Subsequently we will discuss those procedures which are based, more or less, on Cattell’s ramifying linkage analysis (3.10.1 and 3.11) and then a method is treated which only falls in this section because the author is that of the method commented on in 3.10.1.