[R] multivariate local regression with locfit
Paul W Leu
Leu_Paul_W at cat.com
Tue Jun 19 15:59:27 CEST 2001
I've been trying to run locfit on data with 6 inputs and 1 output in R.
Whenever I make a prediction for the same exact data that the model was
built on though, I get significant discrepancies between the fitted outputs
of the prediction and the actual data.
Here is an example of some of the parameters I tried:
fit
<-locfit(plossactual~Reynolds.Bevel+Reynolds.Pinion+Pinion.Velocity+Bevel.Velocity,data=nte,ev
"x", alpha=c(0, .035), scale=c(2e6, 6e5, .6, 1.2, 200, 200))
As you can see, the bandwidth is an extremely small constant value and so
all data should be fitted close to exact (I tried even smaller too). The
variable "x" contains exactly the same data that the model is using so
there shouldn't be any issues of interpolation. When I used the predict
function, I used the same set of data ("x" and "predictionpoints" are the
same). I also tried a variety of scaling schemes, including just setting
the value to T.
Currently, I am just trying to verify that the locfit algorithm does
in fact work for multivariate data by building a model on some data and
then using the model to predict the same set of data, before I start
getting into issues with interpolation, extrapolation, cross validation,
etc. Am I missing something here? Thanks for your time and your help.
Sincerely,
Paul Leu
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
More information about the R-help
mailing list