On 26/02/14 11:57, varin sacha wrote:
Hi,
I have realized a multiple linear regression.
To know how well my model does in terms of prediction, I can compute prediction
intervals bands and decide if they are narrow enough to be of use. If they are
too wide, then they probably are not useful.
So what I am trying to do is :
Theoretically I know that I can use the "predict" command in R to generate the
prediction interval for a set of points. The idea is to find the linear regression using
the lm command. Then I can use the predict command to get the prediction interval for a
set of points in the domain. Then I plot out the predicted values as well as the upper
and lower limits of the prediction intervals for those values.
My problem is to practice what I theoretically know, especially using R.
My linear model is the following :
LinearModel.1 <- lm(GDP.per.head ~ Competitivness.score + Quality.score,
data=Dataset)
summary(LinearModel.1)
predict(LinearModel.1, se.fit = FALSE, scale = NULL, df = Inf,interval = c("none", "confidence",
"prediction"),level = 0.95, type = c("response", "terms"),terms = NULL)
Could you please help me with my R codes ?
If you want prediction intervals, ask for them!!! There is no point at
all in simply repeating the default for the argument "interval". If you
actually want the default, then don't say anything at all. I.e. omit
the "interval" argument from your call and don't clutter things up with
irrelevancies.
But here you *don't* want the default (which is "none"). You want
"prediction". So set interval="prediction" in your call. Actually you
can just set interval="p" because of partial argument matching.
cheers,
Rolf Turner
______________________________________________
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.
