Download ebooks file Spatial Cluster Modelling Monographs on Statistics and Applied Probability 1st Edition Andrew B. Lawson all chapters
Download ebooks file Spatial Cluster Modelling Monographs on Statistics and Applied Probability 1st Edition Andrew B. Lawson all chapters
Download ebooks file Spatial Cluster Modelling Monographs on Statistics and Applied Probability 1st Edition Andrew B. Lawson all chapters
com
https://ebookname.com/product/spatial-cluster-modelling-
monographs-on-statistics-and-applied-probability-1st-
edition-andrew-b-lawson/
OR CLICK BUTTON
DOWNLOAD EBOOK
https://ebookname.com/product/cluster-analysis-fifth-edition-wiley-
series-in-probability-and-statistics-brian-s-everitt/
ebookname.com
https://ebookname.com/product/an-introductory-guide-to-disease-
mapping-1st-edition-andrew-b-lawson/
ebookname.com
https://ebookname.com/product/statistical-analysis-and-modelling-of-
spatial-point-patterns-statistics-in-practice-1st-edition-janine-
illian/
ebookname.com
https://ebookname.com/product/stoppard-s-theatre-finding-order-amid-
chaos-literary-modernism-series-1st-edition-john-fleming/
ebookname.com
100 People Who Are Screwing Up America First Edition
Bernard Goldberg
https://ebookname.com/product/100-people-who-are-screwing-up-america-
first-edition-bernard-goldberg/
ebookname.com
https://ebookname.com/product/memory-maps-the-state-and-manchuria-in-
postwar-japan-the-world-of-east-asia-mariko-asano-tamanoi/
ebookname.com
https://ebookname.com/product/the-dynamics-of-managing-diversity-
second-edition-gill-kirton/
ebookname.com
https://ebookname.com/product/man-meets-dog-routledge-classics-2nd-
edition-konrad-lorenz/
ebookname.com
https://ebookname.com/product/cutaneous-lymphomas-basic-and-clinical-
dermatology-1st-edition-gunter-burg-editor/
ebookname.com
Anti IgE Therapy in Asthma and Allergy 1st Edition S.H.
Arshad
https://ebookname.com/product/anti-ige-therapy-in-asthma-and-
allergy-1st-edition-s-h-arshad/
ebookname.com
SPATIAL
CLUSTER
MODELLING
SPATIAL
CLUSTER
MODELLING
Edited by
Andrew B. Lawson
Department of Mathematical Sciences
University of Aberdeen
Aberdeen, UK
This book contains information obtained from authentic and highly regarded sources. Reprinted material
is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable
efforts have been made to publish reliable data and information, but the authors and the publisher cannot
assume responsibility for the validity of all materials or for the consequences of their use.
Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic
or mechanical, including photocopying, microfilming, and recording, or by any information storage or
retrieval system, without prior permission in writing from the publisher.
All rights reserved. Authorization to photocopy items for internal or personal use, or the personal or
internal use of specific clients, may be granted by CRC Press LLC, provided that $1.50 per page
photocopied is paid directly to Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923
USA. The fee code for users of the Transactional Reporting Service is ISBN 1-58488-266-
2/02/$0.00+$1.50. The fee is subject to change without notice. For organizations that have been granted
a photocopy license by the CCC, a separate system of payment has been arranged.
The consent of CRC Press LLC does not extend to copying for general distribution, for promotion, for
creating new works, or for resale. Specific permission must be obtained in writing from CRC Press LLC
for such copying.
Direct all inquiries to CRC Press LLC, 2000 N.W. Corporate Blvd., Boca Raton, Florida 33431.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation, without intent to infringe.
List of Contributors xi
Preface xiii
3.7 Prediction 54
3.7.1 Conditional Simulation for Neyman-Scott Processes 55
3.7.2 Conditional Simulation for LGCPs 55
3.7.3 Conditional Simulation for Shot-noise G Cox Pro-
cesses 56
3.8 Discussion 58
References 259
Index 277
1.1 Introduction
200
•
•
•
•
•• •• • • •• • •• • ••• •• ••
•
•
•
•
• • •
•
•
•
• •
•• •
•• •• •
4.28 •• • • •• •
•• • • •• •
•
• •
•
•
•
••
•• • •• •• • •• ••
•• • •
•
•• •
• •
•
• • •
•• • •• •• • •• • • ••
• • •
•
•• •• • • • • •
•
•
•
• ••
•••• • • • •
4.26 •
• •• •• •• • • ••• • •
•
150
• •
• •
• •
• •
•
••
•
•• •• • ••
• •• •• • • • • • • • •
• • •
•• ••
• ••
•
•
4.24 • • ••• •• •
•
•
•
• • •
• • •
•• • • •• • •
• •
• ••
• •• •••• •• • • •• • • •
• • •
• • •
• • •• •
•• •
•• •• •• •• •
• • • • • • • •
•
y-coord
•
4.22
•• • ••
•
•• • •• •• • • •
100
• • • • ••
• • • • •
•
•• • • • • • • • • •• • ••
•
• •
• • • •
• •
•• •
4.2
•• • • • • ••• • ••• •• • •
•••• • • ••• • • •• •
•
• • • •• •
• • •
•
•••••
• • •
• •
• •
• • •
•
•
•• • • • • • ••
•
•
• • •
4.18 • • • • •
•
• •• • • • • • • •• •• •
• •
•• • •
• • • •
• • •
50
•• • ••
•
•• • • •
•
• •
•• • • •
•
• • • ••
• • • ••
• •
4.16 • • •
•
•
• • •• • •• ••
•
• •
•
•• •
•
• • •
•• • • • • •••
• • •
• • ••
•
4.14
•
•
••
• •• • •
• • •
•••• ••• • • •
••
• •
•• •
•
•
•• • • • •
•
• ••
• •
•• •••••
•
• • • • •
• •
• • •• •• • •• • • •
• •
• •
•
• •
•
0
4.12
3.48 3.5 3.52 3.54 3.56 3.58 3.6 3.62 0 20 40 60 80 100
4
x 10 x-coord
a function of distance from the cluster centre set within an infinite plane.
The appearance of such clustered event sets is controlled by the location
of centres and the spread of the distribution around these centres.
The dichotomy between these two approaches leads from the different
focus and objectives of the analyses. While both approaches use global
parameters to describe the local variation in space, they are distinguished
by their specificity with regard to cluster location. Within this volume a
range of approaches to clustering is described including examples from both
types of analysis and some where the distinctions are more blurred.
Most of the earliest work in clustering related to the development of
theoretical models for clusters and cluster processes, with limited applica-
tion of the models to real data. In addition, the earliest theoretical work
concerned one-dimensional processes. Early reference sources for such mod-
els and their properties are Snyder (1975) and Cox and Isham (1980). In
the 1970s, increased attention was paid to the application of cluster mod-
els to ecological and biological examples. In these developments, summary
statistics were developed which would allow the assessment of first second
order properties of realisations of point processes. The Ripley-K function,
nearest neighbour and inter-event statistics are all examples of these devel-
opments (see Diggle 1983). Such methods have been further developed in
applications to processes of objects other than points (e.g. Baddeley 1999).
During the 1980s, developments in the analysis of pixellated images led
to the application of Markov random field (MRF) models. These models
were intended for segmentation or noise reduction in images. These models
were often characterised by the inclusion of a neighbourhood effect, often
included within a prior distribution for the image. Bayesian hierarchical
modelling of images was a natural development as prior beliefs about the
spatial structure of the image could be incorporated within a spatial prior
distribution, while measurement error could be included within the likeli-
hood.
Associated with these developments, were the increased use of stochastic
algorithms for maximising or sampling posterior distributions of image fea-
tures – in particular, the development of Gibbs and Metropolis-Hastings
samplers (e.g. Gilks et al. 1996) and simulated annealing for maximisa-
tion (Ripley 1988). These developments regard aggregation as described by
non-specific neighbourhood structures with associated global parameters.
As such they have similarities to geostatistical models where continuous
variation is modelled by predefined covariance structures. Essentially these
are random effect models.
At the other end of the spectrum during the 1980s and 1990s, another
form of image modelling developed which emphasised the specific locational
features of the image. This approach can be described as object recognition
or object modelling. In this collection of approaches the locational charac-
teristics of the image features are specifically modelled (e.g. the landmark
HISTORICAL DEVELOPMENT 5
methods of Dryden and Mardia 1997). Object process modelling is one area
which has seen considerable development. Here a noisy image is assumed
to have an underlying distribution of objects. The aim is to reconstruct
the noise-free object set. Often for simplicity the objects are assumed to
be simple geometric shapes and have associated with them a specific point
which fixes their location. Simple processes could consist of lines, circles or
even tessellations and triangulations.
The recovery of such processes from images has seen the development
and application of sophisticated Markov Chain Monte Carlo (MCMC)
(Baddeley and van Lieshout 1993, Cressie and Lawson 2000) culminat-
ing in the use of special birth-death MCMC algorithms (Geyer and Møller
1994, Stephens 2000).
The above two approaches to image analysis have many variants which
consist of either components of random effect and object recognition tasks.
For example some models attempt to group or classify areas of images, and
these partition models can be thought of as cluster models with locational
components. A review of these different approaches is found in Hurn et al.
(2001).
While image analysis has seen the greatest development of novel method-
ologies, other application areas have also seen considerable progress in re-
cent years, particularly in the application of Bayesian hierarchical models.
Such areas as climatology, geosciences, environmetrics, genetics and spatial
epidemiology have seen considerable advances in the modelling of clustered
data. Within climatology and geo-/environmental sciences the development
of geostatistical methods related to kriging and Bayesian hierarchical mod-
els for random effects has been marked (e.g. Wikle et al. 1998, Mardia et
al. 1998).
In the last decade a considerable body of work has developed with a
focus on small area health data, mainly in the area of spatial epidemiology.
The study of disease clustering has a considerable history and many ap-
proaches have been advanced ranging from hypothesis testing for clusters
to more recently non-specific (random effect) and specific cluster modelling
(for a review see Chapter 6 of Lawson 2001). Model-based approaches to
clustering of diseases has developed both in terms of random effect mod-
elling of clustering (Clayton and Bernardinelli 1992, Breslow and Clayton
1993, Bernardinelli et al. 1995, Waller et al. 1997a, Zia et al. 1997, Knorr-
Held and Besag 1998, Langford et al. 1999, Diggle et al. 1998, Ghosh et
al. 1998, Best et al. 1998, Byers and Besag 2000, Knorr-Held 2000, Sun et
al. 2000, Best and Wakefield 1999, Lawson and Clark 2001) and, to a lesser
extent, in specific modelling of clusters and their locations (Lawson and
Clark 1999a, Lawson and Clark 1999b, Gangnon and Clayton 2000). Vari-
ants of these two extremes of random effect and random object modelling
have been proposed. For example, partition models which group small ar-
eas into constant risk classes have been suggested as a form of cluster
6 SPATIAL CLUSTER MODELLING: AN OVERVIEW
Usually the focus is on estimation of the feature vector θ. For example, these
could be means and variances of Gaussian densities, or could be other rel-
evant feature parameters. If classification is the goal, then once θ is known
the mixture can be decomposed into its components and a maximum a
8 SPATIAL CLUSTER MODELLING: AN OVERVIEW
to
y = z(x, t) + ε(x, t)
a classification likelihood
n
Lclass = fγi (xi |θγi )
i=1