Message-ID: <443E6DE2.9070608@seu.edu.cn>
Date: 2006-04-13T15:27:30Z
From: Xinan Yang
Subject: [Bioc-devel] MCRestimate 1.2.6, stratify bug
In-Reply-To: <443CEC10.8060509@dkfz-heidelberg.de>
Thanks Markus,
I update the related packages, and it works now.
xinan
Markus Ruschhaupt wrote:
> Hi,
>
>
>> I use MCRestimate 1.2.6,
>> but failed to set the parameter "stratify=TRUE", which call
>> balanced.folds to cut the samples into folds.
>>
>> the samples I used are 28 in classed and 29 in another class.
>>
>> It is as below:
>>
>> res.cv <- MCRestimate(eset,
>> + class.column="class",
>> + variableSel.fun = "varSel.highest.t.stat",
>> + classification.fun ="SVM.wrap",
>> + poss.parameters=list(kernel="linear"),
>> + cross.outer=10,
>> + cross.repeat=2,
>> + cross.inner=5,
>> + stratify=TRUE)
>> 12Error in sampleOfFolds[[sample]] : subscript out of bounds
>> >
>>
>>
>> Anyone can give me a hint?
>
>
> I tried to reproduce your error and created a random data set with 28
> samples in one group and 29 samples in the other group. But in my case
> everything worked fine. So first you might try to update all the
> packages you use (especially the pamr and the e1071 package).
>
> Best
> Markus
>
>