Adds parametric interval methods for AFT models ala survreg.
I have tested these functions against the datasets included in the survival package, and my simulations are available on IDA bitbucket.
The intervals perform well under a mild to moderate amount of right censoring. Have not extensively tested left or intervals censoring schemes. In fact, I'm kind of floored by how close the empirical coverage probs are to nominal levels in (essentially) all simulations. This is a really a feather in the Delta method's cap.
There are references to bootstrap methods in each of the files, but I'm only confident in the boot method for add.pi.survreg. The other methods are quite fragile; and because I found such a small difference between the parametric methods and the bootstrap methods, bootstrap is disabled in add.{ci/probs/quantile}.survreg.
Feel free to comment (or not) while I do some more testing. Currently, this branch of ciTools is passing a CRAN built check.
Adds parametric interval methods for AFT models ala survreg.
I have tested these functions against the datasets included in the survival package, and my simulations are available on IDA bitbucket.
The intervals perform well under a mild to moderate amount of right censoring. Have not extensively tested left or intervals censoring schemes. In fact, I'm kind of floored by how close the empirical coverage probs are to nominal levels in (essentially) all simulations. This is a really a feather in the Delta method's cap.
There are references to bootstrap methods in each of the files, but I'm only confident in the boot method for add.pi.survreg. The other methods are quite fragile; and because I found such a small difference between the parametric methods and the bootstrap methods, bootstrap is disabled in add.{ci/probs/quantile}.survreg.
Feel free to comment (or not) while I do some more testing. Currently, this branch of ciTools is passing a CRAN built check.
Thanks.