CSQ - Computational Statistics Quarterly, Vol. 2, Issue 1, 1985, page 125 - 134

by David Wishart

This paper describes an expert system for classifying cases by reference to a
hierarchical cluster analysis on a training set. It tackles the practical
problems encountered in the treatment of mixed types of data with missing
values, and allows variable weights and case weights to be specified.

Two new procedures have been implemented in the computer package Clustan. The
"Cluster" procedure obtains a hierarchical cluster analysis by 4 methods, with
a choice of 13 similarity criteria. Results are stored in a data base for use
with the "Classify" procedure, an interactive identification system which
obtains provisional class and nearest neighbour analyses on trial cases by a
fast tree traversal algorithm. Hissing values can be imputed by analogy wiih
observations on nearest neighbours or clusters.

This document has recently been digitised and made available here. Digitising our collections requires significant resources. Please consider making a donation to the Centre if you download this document!  The donate button is at the top of this page, by the search bar.

The following document were found together:

• Clustan - David Wishart's letter to Dr M A Whyte, Dept of Geology, The University of Sheffield
• Clustan - Design of a Command Language for Interactive or Batch Use in a Statistical Package
• Clustan - Estimation of Missing Values and Diagnosis Using Hierarchical Classifications
• Clustan - Introduction to Using Clustan on the IBM 3083
• Clustan User Manual - Cluster analysis software - Fourth Edition

Reference Number : CSQ_2_1

Date Published : May 1985

Manufacturer : Physica-Verlag, Vienna, Austria

Platform : Clustan

Format : PDF (5 pages)