I have switched from The Unscrambler to R for pls regression analysis and
have been able to calculate scores, coefficients, RMSEP from a large number
of PLS1 and PLS2 models. The ultimate goal is to use these models for
predicting unknown samples, which again is straight-forward with the
built-in predict() function. However, I?m struggling with prediction
uncertainty (i.e. confidence intervals) on predicted values (as an estimate
on the reliability of the predicted values).
Has anyone looked into and/or developed an algorithm or function that
calculates the prediction uncertainty? In order to report on the accuracy
and reliability of the predicted values, we need to report on the yDeviation
(as in
http://www.camo.com/TheUnscrambler/Appendices/The%20Unscrambler%20Method%20References.pdf
on page 31). I have extensively read and searched the available literature
on plsr, mvr, predict, etc. as well as the Nabble forums but I couldn't find
any reference to this kind of uncertainty values.
I am considering writing my own function for this, but if this has already
been addressed, it would be most helpful and would save me a lot of time.
Thanks,
Lieve Laurens, PhD
National Renewable Energy Laboratory
Golden, CO 80401
--
View this message in context:
http://www.nabble.com/Confidence-intervals-PLS-prediction-tp25691532p25691532.html
Sent from the R help mailing list archive at Nabble.com.
