American Statistical Association New York City Metropolitan Area Chapter Mailman School of Public Health Columbia University Department of Biostatistics Colloquium

High-dimensional data have recently arisen in several scientific areas, such as genomic and proteomic studies. Variable selection in such high-dimensional data problems has increasingly become an important problem. Several regularized regression methods have been proposed to allow for jointly selecting variables and estimating regression coefficients. However, inference procedures in such regularized regression methods have not been well developed. We propose several inference procedures for variable selection methods. The proposed methods are evaluated using simulation studies and illustrated using a HIV data set.