R code for RF on German credit data

##The German Credit data

##install.packages("randomForest");

library(randomForest);

x <- read.table("germanCreditNum.Data",header=FALSE,sep="");

n <- nrow(x); p<-ncol(x);

x[,p]<-as.factor(x[,p]);

set.seed(111);

idx2<-sample(1:n,floor(n/2), replace=FALSE);

xtrain<-x[idx2,1:(p-1)]; y<-x[idx2,p];

xtest<-x[-idx2,1:(p-1)]; ytest<-x[-idx2,p];

set.seed(111);

MTRY<-floor(2*sqrt(p-1)); NTREE<-50;

nc<-2; cutoff<-rep(1/nc,nc);

myrf<-randomForest(xtrain,y,xtest=xtest,ytest=ytest,

ntree=NTREE, mtry=MTRY,

replace=TRUE,classwt=NULL,cutoff,strata,

sampsize=nrow(xtrain),

nodesize=if(!is.null(y) && !is.factor(y)) 5 else 1,

importance=FALSE, localImp=FALSE,nPerm=1,

proximity=FALSE,oob.prox=FALSE,

norm.votes=TRUE,do.trace=FALSE,

keep.forest=!is.null(y) && is.null(xtest),corr.bias=FALSE,

keep.onbag=FALSE);

print(myrf);