Seminar at SMU Delhi
April 10, 2013 (Wednesday) ,
3:30 PM at Webinar
Indian Statistical Institute, Delhi
Inference using Adaptive Lasso based residuals
Abstract of Talk
We study the properties of the Adaptive Lasso based residuals in an increasing dimensional linear regression model. It is shown that under suitable sparsity assumptions, the Adaptive Lasso based residual empirical process satisfies a functional oracle property. We use this feature to construct confidence bands for the underlying error distribution function and prediction intervals for unobserved responses, in cases where traditional methods are known to fail.