A0541
Title: A new spatial scan statistic for multiple spatial clusters
Authors: Mohamed Salem Ahmed - University of Lille (France) [presenting]
Michael Genin - University of Lille (France)
Matthieu Marbac - CREST - ENSAI (France)
Abstract: The focus is on the development of a spatial scan statistic able to detect multiple spatial clusters and test their significance, as well as ensure a reasonable computation time for large spatial data. The proposed method is based on generalized linear models in which the spatial clusters are integrated assuming that they have arbitrary parametric shapes (allowing elliptical and rectangular cluster shapes). This allows detecting spatial clusters by estimating their parametric shapes rather than proceeding with an exhaustive search over a set of candidate clusters. We propose a new Monte-Carlo procedure to evaluate the statistical significance of the detected spatial clusters and to estimate the actual number of spatial clusters. Simulations and a case study show that the proposed method is able to consistently and efficiently detect multiple spatial clusters.
