Re: [Wekalist] (no subject)

[Eibe F. <ei...@wa...>]

> On 19/04/2016, at 6:06 AM, Amir Hadanny <amir.had(a)gmail.com> wrote:
> 
> I'm trying to compare the predictions of 7 models using the model 10 selected attributes.
> 
> followed Eibe advice and used the Attributeselectedclassifier in the experimenter - 10cv fold per each, and 10 iterations of a dataset of 2782 patients, 54 attributes.
> 
> (see scrnshot for one of the models out of the 7).
> it takes around 12-20 hours to run...at first it crashed with out of heap - enlarged it to 2048....and crashed again.
> 
> What am i doing wrong? why does it take so long?the running of the attributeselectedclassifier in the explorer for 1 model takes less than 10 seconds...

That’s strange. Does it take less than 10 seconds for all 7 schemes in the Explorer?

Cheers,
Eibe