7.1 Module Clusfind

This module comprises the following six algorithms:

Algorithm

Description of the Algorithm

Data Input

PAM

Partitioning around medoids

A crisp partitioning method for smaller data sets

n n dissimilarity matrix

n p data matrix ( interval scale variables)

CLARA

Clustering large applications

Partitioning method for large data sets

( More than 100 objects)

n n dissimilarity matrix

n p data matrix ( interval scale variables)

FANNY

Fuzzy analysis

Fuzzy partitioning method based on the concept of membership

n n dissimilarity matrix

n p data matrix ( interval scale variables)

AGNES

Agglomerative nesting

Agglomerative hierarchical clustering method

n n dissimilarity matrix

n p data matrix ( interval scale variables)

DIANA

Divisive Analysis

Divisive hierarchical clustering method

n n dissimilarity matrix

n p data matrix ( interval scale variables)

MONA

Monothetic Analysis

Divisive hierarchical clustering method

 

n p data matrix (Binary variables)

If the input matrix is a correlation matrix with elements rij , the elements of the dissimilarity matrix are calculated, using one of the two formulae:

SIGN

d ij = (1- r ij)/2

ABSOLUTE

d ij = 1- r ij