Join us at 3 PM on Friday to hear Dr. Watagoda speak on...
Inference after Variable Selection using Prediction Region Method
Abstract: This talk will introduce three new confidence regions developed using prediction regions. We will be converting prediction regions to confidence regions using bootstrap samples. We named this process, “the prediction region method”. Inference was done for the multiple linear regression model after variable selection using the prediction region method. Theory suggests that the three confidence regions tend to have coverage at least as high as the nominal coverage if the sample size is large enough. We will look at the simulation setup and the results. During the last part of the talk, we will go through a real data example in R.
Lasanthi the department as a Visiting Assistant Professor in August 2017 and was later hired as Assistant Professor. Her research interests lie in the fields of Linear Regression, Variable Selection, Prediction Intervals and the Bootstrap. She works with the Actuarial Science program as well.