[R] Questions regarding MCRestimate package
stellas
stellalsyr at hotmail.com
Wed Aug 1 20:22:06 CEST 2012
Hello,
I'm currently using MCRestimate package and I have a question regarding
the MCRestimate function.
Here is my code:
NestedCV.rf<-MCRestimate(eset, "Class", classificatin.fun="RF.wrap",
variableSel.fun="varSel.highest.var", poss.parameters=
list(var.numbers=c(100), mtry=c(10,50),
cross.outer=10,cross.inner=10,cross.repeat=3)
I'm pretty sure that I was providing "eset" and "Class" correctly. But I
was always getting a warning message:
InMCRestimate.default(eset,"Class",classification.fun="RF.wrap",: There
may be at least one parameter in the parameter list that does not
correspond to a parameter of the classification or any preprocessing
function.
When I removed "mtry=c(10,50)", it worked well without giving any warning
message. So it seems that it was the "mtry" that led to that warning
message. However, I found the information on Bioconductor Code
Search:Template.R/MCRestimate/inst/scripts/Template.R, see following:
Package: MCRestimate - Version: 2.13.0 - Language: r
library(MCRestimate)
library(randomForest)
library(pamr)
library(e1071)
the.expression.set <- get(load(DataSet))
if(RF){
list.of.parameter <-
c(list(mtry=mtry.range,ntree=ntree.range),parameter.for.preprocessing)
r.forest <- MCRestimate(the.expression.set,
class.colum,
classification.fun="RF.wrap",
poss.parameters=list.of.parameter,
thePreprocessingMethods=thePreprocessingfunktionsRF,
cross.outer=cross.outer,
cross.inner=cross.inner,
cross.repeat=cross.repeat,
reference.class=ref.class,
plot.label=plot.label,
rand=SEED)
save(r.forest, file=paste("backRF",SEED,".RData",sep=""))
}
So it seems that specifying parameters(eg. mtry..) for RF.wrap is allowed.
Could anyone please help me figure out what's going on here? Thanks very
much.
Sincerely,
Stella
--
View this message in context: http://r.789695.n4.nabble.com/Questions-regarding-MCRestimate-package-tp4638732.html
Sent from the R help mailing list archive at Nabble.com.
More information about the R-help
mailing list