Report：Partition-based ultrahigh-dimensional variable screening

Time：3:30pm  June20,2017

Place：Room102 Building No.9

Reporter：Jian Kang is an Assistant Professor in the Department of Biostatistics and is a faculty member of the Kidney Epidemiology and Cost Center (KECC) at the University of Michigan.  He received his Ph.D. in Biostatistics from the University of Michigan in 2011. He was an Assistant Professor in the Department of Biostatistics and Bioinformatics and the Department of Radiology and Imaging Sciences at Emory University from 2011 - 2015. He was a core faculty member in the Center for Biomedical Imaging Statistics (CBIS) at Emory University.  His primary research interests are in developing statistical methods for large-scale complex biomedical data with application in precision medicine, imaging, epidemiology and genetics. 

Abstract：Traditional variable selection methods are compromised by overlooking useful information on covariates with similar functionality or spatial proximity, and by treating each covariate independently. Leveraging  prior grouping information on covariates, we propose  partition-based screening methods for ultrahigh-dimensional  variables. We develop these methods under the framework of generalised linear models and show that partition-based screening exhibits the sure screening property with a vanishing false selection rate.  We also propose a data-driven partition screening framework with unavailable or unreliable prior knowledge on  covariate grouping, and investigate its theoretical properties. We consider two special but useful cases: correlation-guided partitioning  and spatial-location-guided partitioning.  In the absence of a single partition, we propose a theoretically-justified strategy for combining  statistics from  various partition methods. The utility of the proposed methods is demonstrated via  simulation and  analysis of functional neuroimaging data.