I do have a two part problem:
a) I am trying to convert C45 .data file to .arff format on a machine using
Ubuntu(Linux). I got the file from KDD archive Forest Cover type data on the
WEKA gives me an error which says: File /home/arthur/desktop/
recognised as an 'C4.5 data files' file.
Reason: nominal value not declared in header: 2596 column 0, read
token, line 1
b) I would like to perform experiments using all-classes-at-once in
My Algorithm is J48.
How can I do this in WEKA?
I am also using this textbook DATA MINING: Practical Machine Learning Tools
and Techniques by Ian H Witten, Eibe Frank and Mark.A.Hall which is very
helpful, but it does not shed in depth light on my queries above.
I am fairly new to WEKA and have very limited Java programming abilities.
I am looking for referral to guiding material:
I have a database of patients data and a binary outcome.
At the moment the DataSet is too small to be entirely classified.
I want to build a classifier that will be built in evolution and based on
repeated users input.
I have two issues I am looking for knowledge about:
1. I want to use an algorithm on my training set but then choose from the
algorithm only the rules that are 100% precise and evaluate them separately
on the test set.
Jrip for example generates a set of rules it considers best and stops.
I want to get a sample of alternative rules (I thought of using bagging for
THEN... I am looking for info about how to get only the rules with precision
of 1 and evaluate them (I am guessing some manipulation of the result
2. The system I want to design is meant to find rules -> present them to a
user -> get feedback about the rule. Any work about such a system I could
So I am looking for guiding material, code snippets, papers or anything you
think might help me.
View this message in context: http://old.nabble.com/Some-novice-help-tp32582825p32582825.html
Sent from the WEKA mailing list archive at Nabble.com.
Thanks Mark, I finally realized I did not have copies of
GenericObjectEditor.props, GenericPropertiesCreator.props and
GUIEditors.props in $HOME/wekafiles/props, so they could not get
loaded. Since I copied them its' all working fine.
On 10/1/2011 11:08 AM, wekalist-request(a)list.scms.waikato.ac.nz wrote:
> Send Wekalist mailing list submissions to
> To subscribe or unsubscribe via the World Wide Web, visit
> or, via email, send a message with subject or body 'help' to
> You can reach the person managing the list at
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Wekalist digest..."
> Today's Topics:
> 1. Weka testimonials (Loic Descotte)
> 2. Complex command line (Nissim Matatov)
> 3. KnowledgeFlow.standardToolBars Not Showing DataSource, Etc.
> ( Christian Straight )
> 4. Re: Complex command line (Harri Saarikoski)
> 5. RE: KnowledgeFlow.standardToolBars Not Showing DataSource,
> Etc. (Mark Hall)
> 6. RE: How one can prove a new algorithm against UCI data
> sets.......... (sudheep)
> 7. Re: Complex command line (Nissim Matatov)
> Message: 1
> Date: Fri, 30 Sep 2011 09:54:54 +0200
> From: Loic Descotte<loic.descotte(a)kelkoo.com>
> Subject: [Wekalist] Weka testimonials
> To: wekalist(a)list.scms.waikato.ac.nz
> Content-Type: text/plain; charset="ISO-8859-1"; format=flowed
> Hello everybody,
> I'm new on this mailing list
> I'm currently evaluating several machine learning tools and I'm looking
> for testimonials about Weka.
> Do you have examples of successful (or famous) applications, web sites,
> companies... that use Weka in production?
> Message: 2
> Date: Sat, 1 Oct 2011 02:45:02 +0300
> From: Nissim Matatov<matatov.n(a)gmail.com>
> Subject: [Wekalist] Complex command line
> To: Wekalist(a)list.scms.waikato.ac.nz
> Content-Type: text/plain; charset="iso-8859-1"
> Hi ,
> I need to run the following command (taken from Explorer's log) :
> weka.classifiers.meta.FilteredClassifier -F
> "weka.filters.unsupervised.attribute.Remove -R 1" -W
> weka.classifiers.meta.CostSensitiveClassifier -- -cost-matrix "[0.0 1.0 1.0;
> 1.0 0.0 1.0; 1.0 1.0 0.0]" -S 1 -W weka.classifiers.trees.J48 -- -C 0.25 -M
> Now I need to supply train and test dataset . Where I need to put -t
> traindataset -T testdataset
> > From weka manual it seems so unclear how to do it
> Thanks a lot
i'm a newbie in WEKA
and want to make some modification on WEKA's ID3
1. i want ID3 to handle numeric data using Usama M. Fayyad and Keki B.
Irani.. where should i do the modification.. like
if (attribute == real)
2. i want ID3 to handle confidence.. well i know where i should put it in
Quinlan's Algorithms but seems i can't find it here..
it should be on bold area
if (allInstance == true)
else if (allInstance == false)
else if (numberofleftattribute == 0)
... do the ID3 thing :)
thank you for the help :)
or even the reading.. :)
View this message in context: http://old.nabble.com/Modification-of-WEKA%27s-ID3-tp32503795p32503795.html
Sent from the WEKA mailing list archive at Nabble.com.
Hi, I am trying to classify a dataset that has 4000 instances with libsvm.
I choose the polynomial kernel. And I got a message around 10-20 minutes
later that says:
Not enough memory.
initial JVM size 57.4
I guess I have to give more memmory to JVM. How can I do that?
Is there a way to increase heap size when i run weka(not for all JVM)?
I need to run the following command (taken from Explorer's log) :
"weka.filters.unsupervised.attribute.Remove -R 1" -W
weka.classifiers.meta.CostSensitiveClassifier -- -cost-matrix "[0.0 1.0 1.0;
1.0 0.0 1.0; 1.0 1.0 0.0]" -S 1 -W weka.classifiers.trees.J48 -- -C 0.25 -M
Now I need to supply train and test dataset . Where I need to put -t
traindataset -T testdataset
>From weka manual it seems so unclear how to do it
Thanks a lot