On Oct 24, 2011, at 2:45 PM, flokke wrote:
Dear all,
I am a R user since about 3 weeks now and still struggeling with
things that
must be very
easy for you...
This week I am struggling with the function predict()
I want to use this function to get a 95% interval.
I understand that you have to use it in such a way as:
lm_examplemodel<- lm(y~x1 + x2 + x3)
Many times the x1,x2,and x3 will be irregularly spaced.
newdata <- data.frame(x1 = ???, x2 = ???, x3 = ???)
You might want to create predictions on a regular grid for plotting
functions htat require such. Obviously you would want to keep the
values within the ranges of original values so you don't extrapolate
beyond the range of measurements.
predict(lm_examplemodel, newdata, interval = 'confidence')
And you might be plotting the estimated and 95% ranges.
And here comes my questions. As you maybe already have guessed its
about
the questions marks/
the values for the new data frame you have to use.
I read in help(predict.lm) that you have to use a new data frame to
be able
to use the predict
function, but I dont know what values I have to assign to x1 and
x2 ... to
create the new data set.
Perhaps:
x1=seq(range(x1)[1]:range(x1)[2], by=diff(range(x1))/10
... und so weiter.
In the help function it says that you can use the fitted values, but
when I
use the function:
fitted(lm_examplemodel)
or fitted.values(lm_examplemodel)
I get many many columns. I think that the function gives a fitted
value for
every Yi and Xi.
That looks wrong. `fitted` should be one per record. You have not
followed the advice to post real data so it is not possible to
diagnose your error. Please read the Posting Guide.
SO how do I get those values? Do i have to use another function than
fitted?
Or not the fitted
values at all?
David Winsemius, MD
West Hartford, CT
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
