摘要:

In this article, we propose an alternative way of testing spatial clustering for common diseases. In order to detect a hot spot, we treat a global cluster statistic from a localized perspective, and define an area with positively correlated neighboring regions as a cluster. The proposed test uses the maximum likelihood method to detect the existence of a cluster, and it does not require the calculation of the mean and variance as most spatial statistic tests do. Using the spatial chi-square test of Rogerson (R) as a benchmark, our subsequent simulations and case study show that when the existence or nonexistence of spatial clusters are apparent, our test result is consistent with R. However, when a low value region surrounded by high value neighbors is considered, the result from R finds a cluster, while our result finds no cluster.

ISSN:1082-4006    

DOI:10.1080/10824009909480526

出版商:Taylor & Francis Group    

年代:1999

数据来源: Taylor

摘要:

Applications in the field of spatial statistics are increasing rapidly: more and more people are interested in incorporating space into their analyses rather than simply dichotomizing or otherwise simplifying the treatment of spatial information. GIS is leading this growth, in the sense thataccessto spatial information is pushing the demand for — and the development of — new techniques. While the major players of GIS (such as Are View and MapInfo) are increasing our access to geo-referenced information, our appreciation of this newfound power must be tempored by the reality that many of the new users don't know much at all about spatial analysis: as Daniel Griffith has said, “… as the issue of computational intensity subsides, and GIS software becomes increasingly user-friendly, more ubiquitously available, and a source for implementing spatial statistical techniques, the danger of malpractice by the non-specialist practitioner grows[23].” We describe in this brief introduction a few of the freely available tools which interested users can count on to help them to get started.

ISSN:1082-4006    

DOI:10.1080/10824009909480527

出版商:Taylor & Francis Group    

年代:1999

数据来源: Taylor