A multivariate nonparametric scan statistic for spatial data

This paper introduces a nonparametric scan method for multivariate data indexed in space. Contrary to many other scan methods, it does not rely on a generalized likelihood ratio but is completely distribution-free as it is based on so-called multivariate ranks. This spatial scan test seems to be more reliable for analysing data that are not Gaussian, such as environmental measurements. We apply this method to a dataset recording the levels of metallic pollutants for two areas in the North of France. (C) 2018 Elsevier B.V. All rights reserved.