(no subject)

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

(no subject)

Sehrish agha
Hello,
I have combined multiple datasets. Now I am using random forest in weka but its not running. I try it with j48 and random tree. Both are working. But random forest is not creating model.
 I am sharing my data please tell me why its not working?

_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to: To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit
https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html

ids 2017 2_weka.arff (7M) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Random forest not working.

Sehrish agha

Hello,
I have combined multiple datasets. Now I am using random forest in weka but its not running. I try it with j48 and random tree. Both are working. But random forest is not creating model.
 I am sharing my data please tell me why its not working?


_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to: To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit
https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html

ids 2017 2_new.arff (7M) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Random forest not working.

Eibe Frank-2
Administrator
RandomForest may run out of memory on your data. The reason is that you have four nominal predictor attributes in your data that each have a large number of values. RandomForest repeatedly applies RandomTree on subsamples of the full data to build an ensemble. When one of those four nominal attributes is selected for splitting the data at the node of a decision tree, this will create a huge number of branches.

Having said this, RandomForest ran successfully on your data when I ran WEKA with 15 GB of heap space.

You could try using the MergeInfrequentNominalValues filter to reduce the number of values in your nominal attributes before you apply RandomForest. A lot of those nominal values only occur once in your data and are unlikely to be useful for prediction anyway.

Cheers,
Eibe

> On 17/11/2019, at 2:55 AM, sehrish Agha <[hidden email]> wrote:
>
>
> Hello,
> I have combined multiple datasets. Now I am using random forest in weka but its not running. I try it with j48 and random tree. Both are working. But random forest is not creating model.
>  I am sharing my data please tell me why its not working?
>
> <ids 2017 2_new.arff>_______________________________________________
> Wekalist mailing list -- [hidden email]
> Send posts to: To unsubscribe send an email to [hidden email]
> To subscribe, unsubscribe, etc., visit
> https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
> List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to: To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit
https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Random forest not working.

Sehrish agha
I applied that filter but still its showing out of memory error. 

On Mon, Nov 18, 2019, 10:00 AM Eibe Frank <[hidden email]> wrote:
RandomForest may run out of memory on your data. The reason is that you have four nominal predictor attributes in your data that each have a large number of values. RandomForest repeatedly applies RandomTree on subsamples of the full data to build an ensemble. When one of those four nominal attributes is selected for splitting the data at the node of a decision tree, this will create a huge number of branches.

Having said this, RandomForest ran successfully on your data when I ran WEKA with 15 GB of heap space.

You could try using the MergeInfrequentNominalValues filter to reduce the number of values in your nominal attributes before you apply RandomForest. A lot of those nominal values only occur once in your data and are unlikely to be useful for prediction anyway.

Cheers,
Eibe

> On 17/11/2019, at 2:55 AM, sehrish Agha <[hidden email]> wrote:
>
>
> Hello,
> I have combined multiple datasets. Now I am using random forest in weka but its not running. I try it with j48 and random tree. Both are working. But random forest is not creating model.
>  I am sharing my data please tell me why its not working?
>
> <ids 2017 2_new.arff>_______________________________________________
> Wekalist mailing list -- [hidden email]
> Send posts to: To unsubscribe send an email to [hidden email]
> To subscribe, unsubscribe, etc., visit
> https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
> List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to: To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit
https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html

_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to: To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit
https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html