Since we want to delete the bootstrap confidence interval from the print and summary functions, it would be nice to have a confint function which gives the bootstrap confidence interval for a specified significance.level.
Currenlty, Georg and Nilufar Akbari from Germany are working on a simulation study where they evaluate estimation of robust confidence intervals in the context of variable selection. Depending on the results, we could then extend this confint function.
The default of confint should give confidence intervals for the global model. "Confidence intervals" for resampled regression coefficients should be available by invoking through a parameter.
Maybe bias-corrected bootstrap CIs could be applied. But first we would have to check if they work in this context (for not selected predictors and the 0).
Since we want to delete the bootstrap confidence interval from the print and summary functions, it would be nice to have a
confint
function which gives the bootstrap confidence interval for a specified significance.level. Currenlty, Georg and Nilufar Akbari from Germany are working on a simulation study where they evaluate estimation of robust confidence intervals in the context of variable selection. Depending on the results, we could then extend thisconfint
function.The default of confint should give confidence intervals for the global model. "Confidence intervals" for resampled regression coefficients should be available by invoking through a parameter.
Maybe bias-corrected bootstrap CIs could be applied. But first we would have to check if they work in this context (for not selected predictors and the 0).