Accession Number:

ADP007174

Title:

A Recursive Partitioning Algorithm for Cluster Analysis,

Descriptive Note:

Corporate Author:

NATIONAL SECURITY AGENCY/CENTRAL SECURITY SERVICE FORT GEORGE G MEADE MD

Personal Author(s):

• Costa, Joseph S., Jr

Report Date:

1992-01-01

Pagination or Media Count:

4.0

Abstract:

In 1965, A.W.F. Edwards and L.L. Cavalli-Sforza introduced a method for cluster analysis based on a recursive partitioning strategy over a minimum-variance clustering criterion. Although this method has been called intuitively appealing, it was dismissed by Gower 1967 and others because of its computational infeasibility. It has been suggested on numerous occasions that some computationally efficient method be found to search an intelligently-chosen subset of the set of all possible partitions for a hopefully near-optimal solution. In this paper, one such method is introduced which borrows from the Classification and Regression Trees CART classification paradigm of Breiman, Friedman, Olshen and Stone 1984.

Descriptors:

• *STATISTICS
• CLASSIFICATION
• CLUSTERING
• PAPER
• STRATEGY

Subject Categories:

• Statistics and Probability

Distribution Statement:

APPROVED FOR PUBLIC RELEASE