If you use return you need to assign the results to an object. For instance
tree.pred <- yourfunction
then tree.pred will contain whatever data you returned from your
function. If you truly want your function to assign objects
internally that are available in your general workspace, instead of
just returning the data you want try:
assign(variablename, datayouwantout, envir=.GlobalEnv)
this will make an assignment to the global environment rather than the
function's environment. In general this seems risky though as your
function could be overwriting data in your main workspace without you
knowing it.
HTH,
Josh
On Wed, May 26, 2010 at 9:26 AM, Changbin Du <changb...@gmail.com> wrote:
> HI, Dear R community,
>
> I am writing the following function to create one data set(*tree.pred*) and
> one vector(*valid.out*) from loops. Later, I want to use the data set from
> this loop to plot curves. I have tried return, list, but I can not use the
> *tree.pred* data and *valid.out* vector.
>
>
> auc.tree<- function(msplit,mbucket) {
>
> * tree.pred<-data.frame()
> valid.out<-vector()*
>
>
> for(i in 1:10) {
> cat('Fold ', i, '\n')
>
> out.fold.c <-((i-1)*c.each.part +1):(i*c.each.part)
> out.fold.n <-((i-1)*n.each.part +1):(i*n.each.part)
>
> train.cv <- n.cc[-out.fold.c, c(2:401, 418)]
> train.nv <- n.nn[-out.fold.n, c(2:401, 418)]
>
> train.v<-rbind(train.cv, train.nv) #training data for feature
> selection
>
> # grow tree
> fit.dimer <- rpart(out ~ ., method="class", data=train.v)
>
> at<-grep("<leaf>", fit.dimer$frame[, "var"], value=FALSE, ignore.case=TRUE)
>
> varr<-as.character(unique(fit.dimer$frame[-at, "var"]))
>
> train.cc <- n.cc[-out.fold.c, ]
> valid.cc <- n.cc[out.fold.c, ]
>
> train.nn <- n.nn[-out.fold.n,]
> valid.nn <- n.nn[out.fold.n,]
>
> train<-rbind(train.cc, train.nn) #training data
> valid<-rbind(valid.cc, valid.nn) # validation data
>
> #creat data set contains the following variables
> myvar<-names(gh5_h) %in% c(varr, "num_cell","num_genes","position",
> "acid_per", "base_per", "charge_per", "hydrophob_per", "polar_per", "out")
>
> train<-train[myvar] # update training set
> valid<-valid[myvar]
>
> control<-rpart.control(xval=10, cp=0.01, minsplit=5, minbucket=5) #control
> the size of the initial tree
> tree.fit <- rpart(out ~ ., method="class", data=train,
> control=control) # model fitting
> p.tree<- prune(tree.fit,
> cp=tree.fit$cptable[which.min(tree.fit$cptable[,"xerror"]),"CP"]) # prune
> the tree
>
>
> #get the prediction for the valid data set.
> tree.pred.r <-predict(p.tree, newdata=valid, type="prob")
> valid.r<-valid$out
>
> tree.pred <-rbind(tree.pred, tree.pred.r)
> valid.out<-c(valid.out, valid.r)
> cat('Dim of tree.pred', dim(tree.pred), 'length of valid.out',
> length(valid.out), '\n' )
>
> }
> *list(tree.pred)
> list(valid.out)*
> cat('Minsplit ', msplit, 'Minbucket', mbucket, '\n')
> cat('10-cross validation is done! \n')
> }
>
>
>> auc.tree(5, 5) #
> Fold 1
> Dim of tree.pred 141 2 length of valid.out 141
> Fold 2
> Dim of tree.pred 282 2 length of valid.out 282
> Fold 3
> Dim of tree.pred 423 2 length of valid.out 423
> Fold 4
> Dim of tree.pred 564 2 length of valid.out 564
> Fold 5
> Dim of tree.pred 705 2 length of valid.out 705
> Fold 6
> Dim of tree.pred 846 2 length of valid.out 846
> Fold 7
> Dim of tree.pred 987 2 length of valid.out 987
> Fold 8
> Dim of tree.pred 1128 2 length of valid.out 1128
> Fold 9
> Dim of tree.pred 1269 2 length of valid.out 1269
> Fold 10
> Dim of tree.pred 1410 2 length of valid.out 1410
> Minsplit 5 Minbucket 5
> 10-cross validation is done!
>
>
> if use return, it will print on the screen, you still can not use it to
> plot. Can anyone help me with this, thanks so much!
>
>
>
> --
> Sincerely,
> Changbin
> --
>
> Changbin Du
> DOE Joint Genome Institute
> Bldg 400 Rm 457
> 2800 Mitchell Dr
> Walnut Creet, CA 94598
> Phone: 925-927-2856
>
> [[alternative HTML version deleted]]
>
> ______________________________________________
> 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.
>
--
Joshua Wiley
Senior in Psychology
University of California, Riverside
http://www.joshuawiley.com/
______________________________________________
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.
