Space-time border analysis to evaluate and detect clusters.

Abstract

This article presents a statistic to identify the proportional intensity of an area to belong to a space-time cluster, which is very important in spatial statistics. A significant extension of the so-called F-function has been described in detail, which, associated with new technology, allows the evaluation of the intensity of change moments in the time process and outline possible occurrences of clusters. The justification of the study is performed through an extensive review of the current literature of the area. The developed algorithms are detailed via computational experiments with simulated data. Real situations are presented and discussed in detail. Finally, topics for future research are indicated.