I was quite thrilled to see how good a job Tibshirani's AVAS function
(additivity and
variance stabilization) does in linearizing the relationship and stabilizing
the variance
for some complex data sets I'm trying to use this method for some prediction
and calibration
problems.
Does anyone have an Splus function that does prediction/calibration
(back-transforming
back to the raw scale) along with standard errors?
I am presently doing calibration (estimating x for a given y) for the AVAS
method in the
following manner (which might be too naive);
1. Predict avas$ty using a CS spline for y v.s. avas$ty,
2. Predict avas$tx using a CS spline of avas$ty v.s. avas$tx, and then
3. Predict x using a CS spline of avas$tx v.s. x.
I can then compute standard errors using bootstrap.
I would appreciate any comments on the above approach, and recommendations on
an alternative approach for prediction/calibration with standard errors using
AVAS.
Thanks in advance,
V. Devanarayan
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news