Spatial cluster detection using nearest neighbor distance
Résumé
Motivated by the analysis of the impact of ecological processes on spatial distribution of tree species, we introduce in this paper a novel approach to detect spatial cluster of points. Our procedure is based on an iterative transformation of the distance between points into a measure of closeness. Our measure has the advantage of being independent of an arbitrary cluster shape and allowing adjustment for covariates. The comparison of the observed measure of closeness to a reference point process leads to a hierarchical clustering of spatial points. The selection of the optimal number of clusters is performed using the Gap statistic. Our procedure is illustrated on a spatial distribution of the Dicorynia guianensis species in the French Guiana terra firme rainforest.
Auteurs, date et publication :
Auteurs Avner Bar-Hen , Mathieu Emily , Nicolas Picard
Publication : Spatial Statistics
Date : 2015
Volume : 14
Pages : 400–411