Print

Print


Statistics seminar
Liverpool University Statistics Division

Time: 2pm, Wednesday May 26th

Venue: Room 6.04, Maths & Oceanography Building, University of Liverpool

Speaker: Helen Wilson, University of Liverpool

Title: A comparison of a parametric and a non-parametric approach to the 
assessment of replicated spatial point patterns

Abstract: The basic methodology for analysing spatial point pattern data is 
now well established (Cressie (1993), Diggle (1983), Ripley (1981)). For the 
single replicate point patterns which dominate the spatial point pattern 
literature, there has been a strong emphasis on fitting parametric models. 
The fields of pathology and materials science, however, present us with 
numerous opportunities for collecting replicated spatial point pattern data. 
As an example, we can easily digitise the positions of cells within multiple 
windows imposed upon brain tissue, using only standard microscopical 
equipment. 
In our work, we exploit replication of this nature, and develop a parametric 
and a non-parametric approach to the assessment of replicated spatial point 
patterns. We begin by proposing a non-parametric method, motivated by analysis 
of variance, which uses the replication as the basis for inference. 
Subsequently, we suggest a parametric, maximum pseudo-likelihood based 
approach to the same problem. 
Specifically, we address the following questions: 
When we specify the correct model for a given data set, is the parametric 
approach more powerful than the non-parametric approach? 
When the model specification is incorrect, does the parametric approach give 
misleading results? 
This is joint work with Peter Diggle (University of Lancaster, U.K.) and Jorge 
Mateu (Universitat Jaume I, E-12071 Castellón, Spain). Microscopy was carried 
out with Dr. Vyvyan Howard (University of Liverpool, U.K.) 



Other forthcoming seminars:

Friday 4 June
Clive Anderson (University of Sheffield)

Tuesday 21 September
Alexandros Karagregoriou (University of Cyprus)
On asymptotic properties of model identification procedures

Full seminar programme available via http://www.liv.ac.uk/maths/SOR/

Damian Clancy
-----------------
[log in to unmask]



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%