ULS Scan Statistic for Hotspot Detection with Continuous Gamma Response

Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the agencies. Abstract: An approach using the upper level set (ULS) scan statistic to detect geospatial hotspots along with its software implementation is presented for continuous response. ULS scan statistic is based on the ULS scan tree. A ULS scan tree is a data structure constructed from response data over a geographic region partitioned into cells. Candidates for hotspots are zones in the region. Each such candidate zone consists of cells that are connected geographically. A ULS scan tree is used to identify candidate zones systematically. Nodes of the ULS scan tree are connected zones. The root (the bottom level) of the ULS scan tree is a zone consisting of the entire region. Zones at the top level (leaf zones) consist of cells with maximal response values. For in between levels, zones at a given level consist of connected cells with higher response values than zones at a lower level. A suitable likelihood statistic and Monte Carlo analysis are used to determine significance of zonal nodes as hotspots. Gamma response model is studied in detail. A case study illustrating application of the gamma response model is presented