M. I. Borrajo García, M. D. Martínez Miranda, W. González Manteiga
In many spatial statistics applications, observations consist of collections of spatial point patterns that may originate from different spatial point processes. Estimating the intensity functions of these processes provides insight into their first-order properties and, in some cases, fully characterizes their spatial distribution. In this talk, we propose a nonparametric clustering method designed to identify clusters of intensity functions. Our approach leverages a multiscale test statistic to detect patterns with common intensities. By incorporating a broad range of bandwidths, it circumvents the challenge of bandwidth selection in testing. To illustrate its practical utility, we apply our method to an environmental dataset, demonstrating its effectiveness in uncovering meaningful spatial structures. We also conduct extensive simulations to evaluate its performance in diverse settings.
Keywords: Point-processes, clustering, first-order intensity
Scheduled
Spatio-Temporal Statistics I
June 11, 2025 3:30 PM
MR 3