Skip to content

Propagate errors with prediction interval

2 messages · Shaunna Morrison, Bert Gunter

#
Hello!


I have a linear dataset with known values of X & Y in R. I treat the
dataset as a linear model.

I have a data point with a known value of X and an unknown value of Y. The
X value has an associated instrument error.

Goal: compute the prediction interval (of Y) for the data point, X, and its
associated error.


Commands:

newdata=data.frame(X=###)

predict(g,newdata,interval="prediction",level=0.68)


This allows me to compute the prediction interval for a data point, based
on a known value of X, but does not propagate the error associated with X.

*I need a method in R to include/propagate the error associated with X. *

Note - The X,Y dataset is subject to the same type of associated error as
the input variable, X. *However*, we do not wish to include the dataset's
associated error in this prediction interval.

Thank you!
#
1. You need to research "errors in variables regression."

2. This is off topic here. This list is about R programming, not
statistics. Post on a statistics list like stats.stackexchange.com
instead.

3. Better yet, consult a local statistical expert.

4. If you post on R issues here, follow the list protocols and post in
plain text, not HTML, please.

Cheers,
Bert
Bert Gunter

"Data is not information. Information is not knowledge. And knowledge
is certainly not wisdom."
   -- Clifford Stoll


On Sun, Jun 28, 2015 at 9:40 AM, Shaunna Morrison
<shaunnamm at email.arizona.edu> wrote: