【Title】Refining Spatial Surveillance with Scan Statistics by Controlling False Discovery Rate
【Speaker】Dr. Xun Xiao, Massey University
【Host】Dr. Yanfu Li
【Time】14:00-15:00, January 17th (Wednesday) , 2018
【Location】 Room N412, Shunde Building
 
【Abstract】Spatial surveillance concerns identifying unusual events spatially scattered over a region with the data aggregated to different sub regions. The scan statistics proposed by Kulldorff in 1997 have become a major tool in Spatial Surveillance. Given all the possible clusters of sub regions, it reports a single cluster maximizing the likelihood ratio as the Most Likely Cluster. Recently, some researchers proposed to detect multiple clusters by controlling the False Discovery Rate. In my talk, I will discuss some improvements of Li’s FDR approach for spatial surveillance. Particularly, if the in-control incidence rate is known, we can use an easier approach to control the FDR of scan statistics. We will also discuss some drawbacks of the FDR approach and other approaches which will be interesting to explore in the future.
 
【Short Bio】Dr. Xun Xiao is a Lecturer in Statistics in Institute of Fundamental Sciences at Massey University, New Zealand. He received his PhD degree from Dept. of Systems Engineering and Engineering Management at City University of Hong Kong in 2016. After his graduation, he conducted postdoctoral studies in City University of Hong Kong and Lulea University of Technology in 2016. His research focuses on the statistical inference of failure time data, degradation modelling, and spatial temporal data analysis.
 
All interested are welcome!