Detecting a change in the interaction parameter
of a spatio-temporal model of Ising-type
Eric Järpe
Proceedings of the 1st Spanish workshop on spatio-temporal modelling
of environmental processes, Universitat Jaume 1, 2001.
Abstract
Surveillance to detect changes of spatial patterns
is of interest in many areas such as environmental control and regional
analysis. a model which possesses both spatial and time dependence is
the Markov chain Markov field. Here a special case of this, called
spatio-temporal interaction model, and change in its parameter
for spatio-temporal interaction is considered. A minimal sufficient
statistic (which is a Markov chain in time) and its asymptotic
distribution are used. A sufficient condition for stationarity
of the sufficient statistic process is given. Conditional
surveillance methods corresponding to common methods for the
time independent case, are derived.