MDL Supervised Discretisation Output

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

MDL Supervised Discretisation Output

Abdrahman0x
Hi all,

I have applied the MDL Supervised Discretisation of Fayyad & Irani available
in Weka. I got in most of the my attributes a result which looks like this:

@attribute attribute2 {'\'All\''}

1. What does it mean?

2. If I have a huge set of attributes, how does the MDL Discretization
method selects the attributes that are going to apply the Entropy and bins.
I know how to compute the Entropy but I need to know the mechanism that weka
use to select which attributes to discretize since i have in my results some
attributes showing {'\'All\''} and seems they are not discritized. Note that
I have read Fayyad and Irani's paper but couldn't realize an answer to my
questions.

Thank you!





--
Sent from: http://weka.8497.n7.nabble.com/
_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: MDL Supervised Discretisation Output

Eibe Frank-3
That means no useful interval boundaries have been found, I.e., the description length is minimised for the “binning” that creates a single bin only. 

Cheers,
Eibe

On Sat, 8 Dec 2018 at 2:06 AM, Abdrahman0x <[hidden email]> wrote:
Hi all,

I have applied the MDL Supervised Discretisation of Fayyad & Irani available
in Weka. I got in most of the my attributes a result which looks like this:

@attribute attribute2 {'\'All\''}

1. What does it mean?

2. If I have a huge set of attributes, how does the MDL Discretization
method selects the attributes that are going to apply the Entropy and bins.
I know how to compute the Entropy but I need to know the mechanism that weka
use to select which attributes to discretize since i have in my results some
attributes showing {'\'All\''} and seems they are not discritized. Note that
I have read Fayyad and Irani's paper but couldn't realize an answer to my
questions.

Thank you!





--
Sent from: http://weka.8497.n7.nabble.com/
_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html

_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: MDL Supervised Discretisation Output

Abdrahman0x
Eibe Frank-3 wrote
> That means no useful interval boundaries have been found, I.e., the
> description length is minimised for the “binning” that creates a single
> bin
> only.

Thank you Eibe for the answer, but I didnt get you well. Based on What there
will be no useful interval boundaries in some attributes, while in the
others will have many. Can you elaborate more please.

Thank you



--
Sent from: http://weka.8497.n7.nabble.com/
_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html