Got it, thanks so much! Greg.
On Mon, Apr 26, 2010 at 11:02 AM, Ridgeway, Greg <gregr@rand.org> wrote:
> Y~X1+X2+X3 is the standard R formula syntax. It simply means "Y is
> predicted by X1 and X2 and X3".
>
> Greg
>
> ------------------------------
> *From:* Changbin Du [mailto:changbind@gmail.com]
> *Sent:* Monday, April 26, 2010 10:21 AM
> *To:* Ridgeway, Greg
> *Cc:* r-help@r-project.org
> *Subject:* Re: R.GBM package
>
> Thanks so much, Greg!
>
> On the demo(bernoulli), I FOUND the following information: IT is used for
> logistic regression.
>
>
> My question is: when I define a decision tree, can I still use the formula
> Y~X1+X2+X3, # formula, even though I dont know the detailed
> formula of decision tree.
>
> Thanks!
>
>
>
>
>
> demo(bernoulli)
> ---- ~~~~~~~~~
>
> Type <Return> to start :
>
> > # LOGISTIC REGRESSION EXAMPLE
> >
> > cat("Running logistic regression example.\n")
> Running logistic regression example.
>
> > # create some data
> > N <- 1000
>
> > X1 <- runif(N)
>
> > X2 <- runif(N)
>
> > X3 <- factor(sample(letters[1:4],N,replace=T))
>
> > mu <- c(-1,0,1,2)[as.numeric(X3)]
>
> > p <- 1/(1+exp(-(sin(3*X1) - 4*X2 + mu)))
>
> > Y <- rbinom(N,1,p)
>
> > # random weights if you want to experiment with them
> > w <- rexp(N)
>
> > w <- N*w/sum(w)
>
> > data <- data.frame(Y=Y,X1=X1,X2=X2,X3=X3)
>
> > # fit initial model
> > gbm1 <- gbm(Y~X1+X2+X3, # formula
> + data=data, # dataset
> + weights=w,
> + var.monotone=c(0,0,0), # -1: monotone decrease, +1:
> monotone increase, 0: no monotone restrictions
> + distribution="bernoulli",
> + n.trees=3000, # number of trees
> + shrinkage=0.001, # shrinkage or learning rate,
> 0.001 to 0.1 usually work
> + interaction.depth=3, # 1: additive model, 2: two-way
> interactions, etc
> + bag.fraction = 0.5, # subsampling fraction, 0.5 is
> probably best
> + train.fraction = 0.5, # fraction of data for training,
> first train.fraction*N used for training
> + cv.folds=5, # do 5-fold cross-validation
> + n.minobsinnode = 10) # minimum total weight needed in
> each node
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> On Mon, Apr 26, 2010 at 9:50 AM, Ridgeway, Greg <gregr@rand.org>
wrote:
>
>> GBM implements boosted trees. It works for 0/1 outcomes, count
outcomes,
>> continuous outcomes and a few others. You do not need to combine rpart
and
>> gbm. You're best bet is to just load the package and run a demo
>> >demo(bernoulli).
>>
>> ------------------------------
>> *From:* Changbin Du [mailto:changbind@gmail.com]
>> *Sent:* Monday, April 26, 2010 9:48 AM
>> *To:* r-help@r-project.org
>> *Cc:* Ridgeway, Greg
>> *Subject:* R.GBM package
>>
>> HI, Dear Greg,
>>
>> I AM A NEW to GBM package. Can boosting decision tree be implemented in
>> 'gbm' package? Or 'gbm' can only be used for
regression?
>>
>> IF can, DO I need to combine the rpart and gbm command?
>>
>> Thanks so much!
>>
>>
>>
>> --
>> Sincerely,
>> Changbin
>> --
>>
>>
>>
>>
__________________________________________________________________________
>>
>> This email message is for the sole use of the intended recipient(s) and
>> may contain confidential information. Any unauthorized review, use,
>> disclosure or distribution is prohibited. If you are not the intended
>> recipient, please contact the sender by reply email and destroy all
copies
>> of the original message.
>>
>>
>
>
> --
> Sincerely,
> Changbin
> --
>
> Changbin Du
> DOE Joint Genome Institute
> Bldg 400 Rm 457
> 2800 Mitchell Dr
> Walnut Creet, CA 94598
> Phone: 925-927-2856
>
>
> __________________________________________________________________________
>
> This email message is for the sole use of the intended...{{dropped:24}}