Cluster Analysis

by Brian S. Everitt; Sabine Landau; Morven Leese

• ISBN13: 9780340761199
• ISBN10: 0340761199
• Edition: 4th
• Format: Hardcover
• Copyright: 2001-07-12
• Publisher: Hodder Education Publishers

Summary

Cluster analysis comprises a range of methods of classifying multivariate data into subgroups, and these techniques are widely applicable. The 4th edition of Cluster Analysis updates the successful 3rd edition and incorporates new material to cover developing areas, such as Bayesianstatistics and neural networks. In addition, recent research on the evaluation of results and the assessment of the number of clusters has also been included. Real-life examples are used throughout to demonstrate the application of the theory, and graphical techniques are demonstrated withappropriate figures. Finally, this edition includes information on the software packages currently available. The author assumes some prior knowledge of statistics, but writes in a non-mathematical, accessible style. This concise book is ideal for postgraduate students of statistics, as well asresearchers in medicine, sociology, and market research.

Author Biography

Brian S. Everitt, Sabine Landau and Morven Leese are all at the Institute of Psychiatry, King's College London.

Table of Contents

Preface	ix
An Introduction to Classification and Clustering	1	(10)
Introduction	1	(1)
Reasons for classifying	2	(2)
Numerical methods of classification---cluster analysis	4	(2)
What is a cluster?	6	(2)
Examples of the use of clustering	8	(2)
Summary	10	(1)
Visualizing Clusters	11	(24)
Introduction	11	(1)
Detecting clusters in one or two dimensions	11	(9)
Visualizing clusters in data sets with more than three variables	20	(10)
Multidimensional scaling	30	(4)
Summary	34	(1)
Measurement of Proximity	35	(20)
Introduction	35	(1)
Similarity measures for categorical data	35	(4)
Dissimilarity and distance measures for continuous data	39	(4)
Similarity measures for data containing both continuous and categorical variables	43	(3)
Inter-group proximity measures	46	(2)
Weighting variables	48	(3)
Standardization	51	(1)
Choice of proximity measure	52	(1)
Missing data values	53	(1)
Summary	54	(1)
Hierarchical Clustering	55	(35)
Introduction	55	(1)
Agglomerative methods	56	(11)
Divisive methods	67	(3)
Applying the hierarchical clustering process	70	(8)
Applications of hierarchical methods	78	(11)
Summary	89	(1)
Optimization Clustering Techniques	90	(28)
Introduction	90	(1)
Clustering criteria derived from the dissimilarity matrix	90	(2)
Clustering criteria derived from continuous data	92	(7)
Optimization algorithms	99	(3)
Choosing the number of clusters	102	(3)
Applications of optimization methods	105	(12)
Summary	117	(1)
Finite Mixture Densities as Models for Cluster Analysis	118	(23)
Introduction	118	(1)
Finite mixture densities	118	(4)
Other finite mixture densities	122	(3)
Tests for the number of components	125	(1)
Applications of finite mixture densities and classification likelihood	126	(14)
Summary	140	(1)
Miscellaneous Clustering Methods	141	(36)
Introduction	141	(1)
Density search clustering techniques	142	(3)
Techniques which allow overlapping clusters	145	(9)
Direct clustering of data matrices	154	(7)
Clustering with constraints	161	(3)
Fuzzy clustering	164	(5)
Clustering and artificial neural networks	169	(7)
Summary	176	(1)
Some Final Comments and Guidelines	177	(20)
Introduction	177	(2)
Using clustering techniques in practice	179	(1)
Testing for absence of structure	180	(1)
Methods for comparing cluster solutions	181	(3)
Internal cluster quality, influence and robustness	184	(4)
Graphical interpretation	188	(2)
Illustrative examples	190	(6)
Summary	196	(1)
Appendix: Software for Cluster Analysis	197	(11)
A.1 Introduction	197	(1)
A.2 Statistical packages incorporating cluster analysis	198	(7)
A.3 Software for mixture modelling	205	(1)
A.4 Software for non-standard clustering methods	206	(2)
Bibliography	208	(23)
Index	231