headers
Mithun Prasad | 1 Dec 2003 02:16
Picon
Favicon

NominalToBinary

Hi,
     I was wondering how to convert nominal attributes into binary numeric 
attributes. An attribute with k values is transformed into k binary 
attributes if the class is nominal... but am not sure how weka does that. 
Thank You.

Cheers,
Mithun.

_________________________________________________________________
Gaming galore at  http://xtramsn.co.nz/gaming !
Permalink | Reply | 0 comments |
headers
Hughes, Ann (CREST | 1 Dec 2003 10:06
Picon

Help using Weka for analysing health data

Hi 
Apologies straight off for the simple questions but I am completely new to
WEKA.

I am completing an MSc in Health Informatics and a thesis is required for
this year, as I said I am completely new to Weka but it was suggested to me
to think about using it to analyse health data and do my thesis on data
mining.

I have access to a large database of patient information relating to cardiac
surgery and I was wondering if it is possible to use Weka in this way? To
clarify - the data relates to pre-surgery risk factors such as diabetes,
hypertension, smoking etc and the surgery details relates to the type of
surgery carried out - length of stay in hospital and complications etc.
There is approx 3 years data available.

I have read a little on Weka and havn't got to grips with it yet as regards
ARFF files etc - on reading the chapter on creating these files it states to
load the csf files into word - how is this done?

I would appreciate any information/help to assist me,
Thank You
Kind Regards
Ann

**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Socrates E. Alikhanidi | 1 Dec 2003 12:06
Picon

Cost-sensitive matrix.

    Dear colleagues,

Could you please point me out to an example of the cost-sensitive matrix of misclassification for Weka. Or
send it me please.

The problem is that the textbook's ("Data Mining") description of cost-sensitive matrix on page 280 is not
compatible with the newer Weka (v. 3.2.3), while the distributive does not contain an example of such a file.

Thank you,
----------------------------------
Sokratis ALIKHANIDI, Ph.D.
Department of Knowledge-based Information Engineering
Toyohashi University of Technology
1-1 Hibarigaoka, Tempaku-cho, Toyohashi 441, JAPAN
TEL: 0532-44-6892
FAX: 0532-44-6873
mailto:socrates <at> mis.tutkie.tut.ac.jp
Permalink | Reply | 0 comments |
headers
Kengkaj Sathianpantarit | 1 Dec 2003 15:06
Picon
Favicon

WEKA crash when many attributes

Hello,
I use Weka 3.4 run Naive Bayes Classifier with about 10,000 attributes
when run about 5 minute program crash (not responding)
 
Hardware
Dual Pentium !!! 933MHz
SD-RAM 512MB
 
OS
Windows 2000 Advanced Server SP3
Java 1.4.1
 
Dataset WebKB 4-University
 
how i fix this problem?
 
Kengkaj, Student at Kasetsart University, Thailand

Do you Yahoo!?
Free Pop-Up Blocker - Get it now 
_______________________________________________
Wekalist mailing list
Wekalist <at> list.scms.waikato.ac.nz
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Permalink | Reply | 0 comments |
headers
Kengkaj Sathianpantarit | 2 Dec 2003 06:10
Picon
Favicon

Neural Network run very long

Hello,
I use NN for classify webpage but it run very long (about 15 hours not yet finish)
What wrong? or by normal NN take much time to run.
 
Kengkaj
Department of Computer Science
Kasetsart University, Thailand
 
 

Do you Yahoo!?
Free Pop-Up Blocker - Get it now 
_______________________________________________
Wekalist mailing list
Wekalist <at> list.scms.waikato.ac.nz
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Permalink | Reply | 0 comments |
headers
Natschläger Thomas | 2 Dec 2003 08:38
Picon
Favicon

BUG? Weka-Experimentor Advanced

Hi!

Today I found that for my system (Weka 3-4, Linux) I consistently get the error "Can not create an example of
weka.experiment.InstanceResultListener form the current classpath" when I'm trying to set up an
experiment in the "Advanced" mode. Note that I've already used the "Simple" mode sucessfully for several experiments.

Here is the shell script I use to start weka:

#!/bin/tcsh
setenv WEKAHOME ~/weka
setenv CLASSPATH ${WEKAHOME}/weka.jar
exec java -Xmx500m -jar ${WEKAHOME}/weka.jar

Here is the output of java -version

java version "1.4.1_02"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.1_02-b06)
Java HotSpot(TM) Client VM (build 1.4.1_02-b06, mixed mode)

Any suggestion for a workaround are welcome.

-thomas
Permalink | Reply | 1 comment |
headers
ozric | 2 Dec 2003 09:48
Picon

Re: WEKA crash when many attributes

Hi, my humble hints are: (1.) Modifiy your memory setting , perhaps java -Xmx400M (2.) Is it really necessary integrate all 10.000 attributes?, if not try the AttributeSelectedClassifier with NaiveBayes. hope this help,christian Hello, I use Weka 3.4 run Naive Bayes Classifier with about 10,000 attributes when run about 5 minute program crash (not responding) Hardware Dual Pentium !!! 933MHz SD-RAM 512MB OS Windows 2000 Advanced Server SP3 Java 1.4.1 Dataset WebKB 4-University how i fix this problem? 

_______________________________________________
Wekalist mailing list
Wekalist <at> list.scms.waikato.ac.nz
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Permalink | Reply | 0 comments |
headers
ozric | 2 Dec 2003 09:54
Picon

Re: Neural Network run very long

What characteristics is in your dataset, so how many instances and attributes? I doing last weeks data-mining with ~ 100.000 train & test instances with 5-10 attributes and recognize that neural-networks and svm are really slowly if you haven't a > 3 GHZ Machine :-) regards,christian Hello, I use NN for classify webpage but it run very long (about 15 hours not yet finish) What wrong? or by normal NN take much time to run. Kengkaj Department of Computer Science Kasetsart University, Thailand 

_______________________________________________
Wekalist mailing list
Wekalist <at> list.scms.waikato.ac.nz
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Permalink | Reply | 0 comments |
headers
Johannes Fuernkranz | 3 Dec 2003 14:09
Picon
Picon

Bug in ClassOrder?

Hi,

I would like to sort class values according to frequency and use 
weka.filters.supervised.attributes.ClassOrder for that. I tried to 
re-use JRip code, but I didn't understand everything. I ended up using 
the following, which seemed to make sense to me:

     // order classes in ascending frequency
     // Instances data  = properly initialized
     ClassOrder m_Filter = new ClassOrder();
     m_Filter.setClassOrder(ClassOrder.FREQ_ASCEND);
     m_Filter.setInputFormat(data);
     data = Filter.useFilter(data, m_Filter);

The problem that I have with that is that 
data.classAttribute().isInRange() works differently after I call that. 
Apparently, the hash table size that is used in the implementation of 
isInRange() changes (e.g., for soybean from 19 to 10). Thus, class 
values with index >= 10 are no longer in range.

Am I doing something wrong?

cheers, Juffi
--

-- 
Johannes Fuernkranz
Austrian Research Inst. for Artificial Intelligence
Permalink | Reply | 0 comments |
headers
Eibe Frank | 5 Dec 2003 04:48
Picon
Picon

Re: BUG? Weka-Experimentor Advanced


I couldn't reproduce the problem. Are you sure you the CLASSPATH is 
correct? It seems like it couldn't find the class file in your 
classpath.

BTW: There is no "InstanceResultListener", only an 
"InstancesResultListener", but I assume that's just a typo.

Cheers,
Eibe

On Wednesday, December 3, 2003, at 12:02  PM, 
wekalist-request <at> list.scms.waikato.ac.nz wrote:

> From: Natschl‰ger Thomas <Thomas.Natschlaeger <at> scch.at>
> Date: Tue Dec 2, 2003  8:38:25  PM Pacific/Auckland
> To: "wekalist (E-mail)" <wekalist <at> list.scms.waikato.ac.nz>
> Subject: [Wekalist] BUG? Weka-Experimentor Advanced
>
>
> Hi!
>
> Today I found that for my system (Weka 3-4, Linux) I consistently get 
> the error "Can not create an example of 
> weka.experiment.InstanceResultListener form the current classpath" 
> when I'm trying to set up an experiment in the "Advanced" mode. Note 
> that I've already used the "Simple" mode sucessfully for several 
> experiments.
>
> Here is the shell script I use to start weka:
>
> #!/bin/tcsh
> setenv WEKAHOME ~/weka
> setenv CLASSPATH ${WEKAHOME}/weka.jar
> exec java -Xmx500m -jar ${WEKAHOME}/weka.jar
>
> Here is the output of java -version
>
> java version "1.4.1_02"
> Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.1_02-b06)
> Java HotSpot(TM) Client VM (build 1.4.1_02-b06, mixed mode)
>
> Any suggestion for a workaround are welcome.
>
> -thomas
Permalink | Reply | 0 comments |

Gmane