Alessia | 1 Mar 09:19 2007
Picon

Re: discretization

excuse me, i don't understand well. I go in that site web and then, how can 
i download the file?

----- Original Message ----- 
From: "Peter Reutemann" <fracpete <at> waikato.ac.nz>
To: "Weka machine learning workbench list." 
<wekalist <at> list.scms.waikato.ac.nz>
Sent: Wednesday, February 28, 2007 9:04 PM
Subject: Re: [Wekalist] discretization

>> do you know where i can find this pubblication?
>>
>>  <at> inproceedings{Fayyad1993,
>>    author = {Usama M. Fayyad and Keki B. Irani},
>>    booktitle = {Thirteenth International Joint Conference on Articial 
>> Intelligence},
>>    pages = {1022-1027},
>>    publisher = {Morgan Kaufmann Publishers},
>>    title = {Multi-interval discretization of continuousvalued attributes 
>> for classification learning},
>>    volume = {2},
>>    year = {1993}
>> }
>
> Just get the proceedings:
> http://www.informatik.uni-trier.de/~ley/db/conf/ijcai/ijcai93.html
>
> Cheers, Peter
> -- 
> Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ
(Continue reading)

Peter Reutemann | 1 Mar 09:42 2007
Picon
Picon

Re: discretization

> excuse me, i don't understand well. I go in that site web and then, how
> can
> i download the file?

I haven't seen the paper available for download anywhere, hence the link
with the detailed information (incl ISBN) about the proceedings to get
them in printed form, from a library for instance.

Cheers, Peter
--

-- 
Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ
http://www.cs.waikato.ac.nz/~fracpete/           Ph. +64 (7) 858-5174
narendra dhulipalla | 1 Mar 12:23 2007
Picon

meaning of result buffer output in weka

Hi All,

Can any one tell me the meaning of rules generated in weka for any classifier?(Result buffer in weka classifiers)
Please try to explain the following example:

J48 pruned tree
------------------

Fault = Policy_Holder
|   PastNumberOfClaims = 2_to_4
|   |   VehicleCategory = Utility
|   |   |   Days:Policy-Accident = 1_to_7: No (1.0)
|   |   |   Days:Policy-Accident = 15_to_30: Yes (0.0)
|   |&n bsp;  |   Days:Policy-Accident = 8_to_15: Yes (4.0)
|   |   |   Days:Policy-Accident = more_than_30
|   |   |   |   NumberOfSuppliments = none
|   |   |   |   |   RepNumber = fourteen: Yes (4.0)
|   |   |   |   |   RepNumber = seven: No (2.0/1.0)
|   |   |   | & nbsp; |   Make = Ford: Yes (0.0)
|   |   |   |   |   Make = VW: Yes (0.0)
|   |   |   |   |   Make = Dodge: Yes (0.0)
|   |   |   |   |   Make = BMW: Yes (0.0)
|   |   |   |   WeekOfMonth = one: Yes (269.0)
|   |   |   |  
WeekOfMonth = one: Yes (283.0)
What does the values in the brackets signify?
In some cases its given only one value in brackets and in other cases its like(2.0/1.0).....?
In some cases the value in the brackets is 0.0?What does this signifies?

If any tutorial is available on the web to explain this , please give me the link of that?



Thanks and Regards
Narendra Dhulipalla
Success is not final, failure is not fatal: it is the courage to continue that counts. --- Winston Churchill
(9948959148)

Finding fabulous fares is fun.
Let Yahoo! FareChase search your favorite travel sites to find flight and hotel bargains.
_______________________________________________
Wekalist mailing list
Wekalist <at> list.scms.waikato.ac.nz
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Pablo Siber | 1 Mar 15:08 2007

Useful tips in text categorization?

Hi all,

I'm a software developer and I've been assigned to a text categorization 
project.
I've been reading the literature so far and I'm aware this is quite common, so 
no need to re-invent the wheel.
For example, this post: 
https://list.scms.waikato.ac.nz/mailman/htdig/wekalist/2003-August/001685.html
([Wekalist] Rookie attempting text categorization, +3 years old) is very 
similar to the job I've got to do.

I've got the impression that:

1. IBk (KNN) algorithms are a good choice because: (i) they can easily add new 
documents in the training set, and (ii) can handle sparse vectors(arff files)

2. The principal "con" here is time for execution, but not such a big deal in 
my case.

Anyone done something similar? 

This project in my shop is "R+D", but anyway we don't have so much time to be 
playing around till figuring stuff out.

I've found the Weka project very interesting and exciting, but probably more 
useful to grad students/computer science researcher/etc than industry.

I'm willing to use the Weka libraries and, with your useful help and insights 
get my project ahead.

Of course, I'm willing to share my experience so it gets documented for future  
projects.

So, anyone with useful tips or experiences?

Thanks in advance, and regards,

P.

_______________________________________________
Wekalist mailing list
Wekalist <at> list.scms.waikato.ac.nz
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Jiang Su | 1 Mar 16:08 2007
Picon

Re: Useful tips in text categorization?

Hi 

> I've got the impression that:
> 
> 1. IBk (KNN) algorithms are a good choice because: (i) they can easily add new 
> documents in the training set, and (ii) can handle sparse vectors(arff files)

Naive bayes multinominal can do the same thing. If you want, I can sent my
implementation
to you.SVM can do the same thing(at least in papers),:). 

BTW, anybody is using SVM to do TC?

Jiang Su
Samuel Robert Reid | 1 Mar 16:34 2007
Picon

RE: Useful tips in text categorization?

Bayesian modeling for text classification has proven to be quite powerful
(see especially the papers at the bottom):
http://psiexp.ss.uci.edu/research/programs_data/toolbox.htm

I saw a c++ implementation of a similar model recently, but can't track it
down now.

Good luck,
Sam

-----Original Message-----
From: wekalist-bounces <at> list.scms.waikato.ac.nz
[mailto:wekalist-bounces <at> list.scms.waikato.ac.nz] On Behalf Of Pablo Siber
Sent: Thursday, March 01, 2007 7:09 AM
To: wekalist <at> list.scms.waikato.ac.nz
Subject: [Wekalist] Useful tips in text categorization?

Hi all,

I'm a software developer and I've been assigned to a text categorization 
project.
I've been reading the literature so far and I'm aware this is quite common,
so 
no need to re-invent the wheel.
For example, this post: 
https://list.scms.waikato.ac.nz/mailman/htdig/wekalist/2003-August/001685.ht
ml
([Wekalist] Rookie attempting text categorization, +3 years old) is very 
similar to the job I've got to do.

I've got the impression that:

1. IBk (KNN) algorithms are a good choice because: (i) they can easily add
new 
documents in the training set, and (ii) can handle sparse vectors(arff
files)

2. The principal "con" here is time for execution, but not such a big deal
in 
my case.

Anyone done something similar? 

This project in my shop is "R+D", but anyway we don't have so much time to
be 
playing around till figuring stuff out.

I've found the Weka project very interesting and exciting, but probably more

useful to grad students/computer science researcher/etc than industry.

I'm willing to use the Weka libraries and, with your useful help and
insights 
get my project ahead.

Of course, I'm willing to share my experience so it gets documented for
future  
projects.

So, anyone with useful tips or experiences?

Thanks in advance, and regards,

P.

_______________________________________________
Wekalist mailing list
Wekalist <at> list.scms.waikato.ac.nz
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Luc Sorel | 1 Mar 16:55 2007
Picon

Re: meaning of result buffer output in weka

narendra dhulipalla a écrit :
Hi All,

Can any one tell me the meaning of rules generated in weka for any classifier?(Result buffer in weka classifiers)
Please try to explain the following example:
Hello,

Have a look there : http://weka.sourceforge.net/wekadoc/index.php/en:Primer (scroll down a bit).

The tree must be read from top to bottom. Each line is a test and the number of "|" before the test accounts for the imbrication of the test. Tests having one "|" more than the test before will be executed only if the output of the test before is "true".

Tests finishing by something like ": No (2, 1)" are leaves (=decisions) of the tree. The numbers in brackets are :
  • first number : number of elements classified at this leaf of the decision tree
  • second number : if any, number of elements classified at this leaf of the decision tree which are misclassified (here, is actually "Yes" but classified as "No")
I hope my explanations are clear.

You can also have a graphical visualisation of your tree, look the same link than above.

Cordially, Luc

J48 pruned tree
------------------

Fault = Policy_Holder
|   PastNumberOfClaims = 2_to_4
|   |   VehicleCategory = Utility
|   |   |   Days:Policy-Accident = 1_to_7: No (1.0)
|   |   |   Days:Policy-Accident = 15_to_30: Yes (0.0)
|   |   |   Days:Policy-Accident = 8_to_15: Yes (4.0)
|   |   |   Days:Policy-Accident = more_than_30
|   |   |   |   NumberOfSuppliments = none
|   |   |   |   |   RepNumber = fourteen: Yes (4.0)
|   |   |   |   |   RepNumber = seven: No (2.0/1.0)
|   |   |   |   |   Make = Ford: Yes (0.0)
|   |   |   |   |   Make = VW: Yes (0.0)
|   |   |   |   |   Make = Dodge: Yes (0.0)
|   |   |   |   |   Make = BMW: Yes (0.0)
|   |   |   |   WeekOfMonth = one: Yes (269.0)
|   |   |   |   WeekOfMonth = one: Yes (283.0)
What does the values in the brackets signify?
In some cases its given only one value in brackets and in other cases its like(2.0/1.0).....?
In some cases the value in the brackets is 0.0?What does this signifies?

If any tutorial is available on the web to explain this , please give me the link of that?



Thanks and Regards
Narendra Dhulipalla
Success is not final, failure is not fatal: it is the courage to continue that counts. --- Winston Churchill
(9948959148)

Finding fabulous fares is fun.
Let Yahoo! FareChase search your favorite travel sites to find flight and hotel bargains. _______________________________________________ Wekalist mailing list Wekalist <at> list.scms.waikato.ac.nz https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
_______________________________________________
Wekalist mailing list
Wekalist <at> list.scms.waikato.ac.nz
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Yohannes Afework | 1 Mar 18:49 2007
Picon

Dataset limit

Hi
I am using Weka-3-5-3 (on a PC with a Windows 2000 OS). I am processing the input data with VB and import it to Weka  in Arff format via MS Excel. But now the number of attributes I have is over the 255 column limit of Excel. What can I do to import the data into Weka?
Thanks
YZA

Everyone is raving about the all-new Yahoo! Mail beta.
_______________________________________________
Wekalist mailing list
Wekalist <at> list.scms.waikato.ac.nz
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Peter Reutemann | 1 Mar 20:47 2007
Picon
Picon

Re: Dataset limit

>   I am using Weka-3-5-3 (on a PC with a Windows 2000 OS). I am processing the input data with VB and import it to
Weka  in Arff format via MS Excel. But now the number of attributes I have is over the 255 column limit of
Excel. What can I do to import the data into Weka?

Weka doesn't have Excel's limit of 255 columns. If you're using VB 
anyway to generate your files, just create CSV files and import/use them 
in Weka. BTW If you upgrade to 3.5.4 or later, you can run, e.g., a 
classifier now from commandline with a CSV file. No need to convert it 
to ARFF first.

The following formats can be used from commandline:
- ARFF (*.arff)
- XRFF (*.xrff) - the new XML-based Weka data format
- CSV (*.csv)
- C4.5 (*.data or *.names)
- libsvm (*.libsvm)
- binary serialized instances (*.bsi)

HTH

Cheers, Peter
--

-- 
Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ
http://www.cs.waikato.ac.nz/~fracpete/           Ph. +64 (7) 858-5174
wohujintao | 2 Mar 02:23 2007

CHAPTER 14:Embedded Machine Learning

Hi,Where can I find more examples as mentioned in Chapter 14? I want to use the algorithms in my own code.
Thanks   

Hi,
Where can I find more examples as mentioned in Chapter 14? I want to use the algorithms in my own code.
 
Thanks
 
 
 


想加入吗?1.9亿用户正在使用网易邮箱 www.126.com
_______________________________________________
Wekalist mailing list
Wekalist <at> list.scms.waikato.ac.nz
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist

Gmane