STATISTICAL ANALYSIS OF MAPPED SPATIAL POINT PATTERNS

No Thumbnail Available
Date
1988-05
Authors
DOGUWA, SANI IBRAHIM
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The object of this study is to examine the various methods which deal with tests of randomness for univariate mapped point patterns. Some of these methods are extended to multitype patterns. Particular attention is paid to the bounded nature of the pattern being studied, and the complications that thereby arise in the implementation of the various methods. Edge corrections are reported for the mean and variance of the average distance between a sampling point and the nearest object, X, for the case of a regular grid of sampling points. Using these corrections the point-object analogue of the Clark-Evans statistic is found to be a powerful detector of departures from randomness in the clustered direction. A new edge-corrected estimator is proposed for the second moment cumulative function K(t), introduced by Ripley for the study of spatial point processes. This new estimator is compared by simulation methods with existing edge-corrected estimators in the context of L(t) function which is used to study point patterns. The result of the simulation study suggests that the new estimator provides almost unbiased estimate of L(t) and has a smaller mean squared error than its predecessors. A new method is proposed for estimating G(w) and F(x), the distribution functions of the object-to-object and point-to-object nearest neighbour distances respectively. The new method makes more complete use of the information available and has a smaller mean squared error than the existing alternatives. The method appears equally effective with random, clustered and regular patterns. Programs for statistical analysis were written in FORTRAN and run on a DEC system - 1.0/99 computer of the University of Essex. The graphs were plotted on the versatec-plotter of the University using SIMPLE PLOT MARK2
Description
A thesis submitted to satisfy the requirements of the degree of Doctor of Philosophy (Ph.D) in the Department of Mathematics, University of Essex.
Keywords
STATISTICAL ANALYSIS,, MAPPED SPATIAL POINT PATTERNS,
Citation
Collections