headers
Sarwat Nizamani | 1 Aug 2010 17:42
Picon
Favicon

RE: Wekalist Digest, Vol 90, Issue 1

Reply to Message 5:
In the original message from I failed means that I did not get the required output attached in the original message.


Best Regards,
Sarwat Nizamani
 



 
> From: wekalist-request <at> list.scms.waikato.ac.nz
> Subject: Wekalist Digest, Vol 90, Issue 1
> To: wekalist <at> list.scms.waikato.ac.nz
> Date: Sat, 31 Jul 2010 10:28:29 -0700
>
> Send Wekalist mailing list submissions to
> wekalist <at> list.scms.waikato.ac.nz
>
> To subscribe or unsubscribe via the World Wide Web, visit
> https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
> or, via email, send a message with subject or body 'help' to
> wekalist-request <at> list.scms.waikato.ac.nz
>
> You can reach the person managing the list at
> wekalist-owner <at> list.scms.waikato.ac.nz
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Wekalist digest..."
>
>
> Today's Topics:
>
> 1. Use NaiveBayesMultinomial classifier with textual datasets
> (Amy Hu)
> 2. Re: help on real value predictions of the number of
> instances (Jing Su)
> 3. Re: one-vs-rest SVM (Mark Hall)
> 4. Re: Function Estimation and Curve Fitting using Weka. (Mark Hall)
> 5. Re: FW: Arff file format (Mark Hall)
> 6. Problem in uploading Supplied Test Set in ver 3.6.3
> (shehan <at> uom.lk)
> 7. extension of weka (m lalshateri)
> 8. what is the unit of time used in weka.experimenter (icu uci)
> 9. Re: extension of weka (Ruchi Rana)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Fri, 30 Jul 2010 13:05:00 +0000
> From: Amy Hu <enjoyrong <at> hotmail.com>
> Subject: [Wekalist] Use NaiveBayesMultinomial classifier with textual
> datasets
> To: <wekalist <at> list.scms.waikato.ac.nz>
> Message-ID: <COL112-W101C5C83E657CCEB9650A9ACAA0 <at> phx.gbl>
> Content-Type: text/plain; charset="gb2312"
>
>
> Hi all,
>
> I am using NaiveBayesMultinomial (NBM) classifier in WEKA with textual datasets. I found that when I use binary representation of documents which is binary presence and absence of words, the performance of NBM is better when using term frequency as document representation. I have thought that term frequency representation might be better for textual data but it seems not here. Has anyone else met this before? Why it is like this? Thanks.
>
>
>
> Regards
>
>
>
> Rong
>
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: https://list.scms.waikato.ac.nz/pipermail/wekalist/attachments/20100730/dbfa1f2a/attachment-0001.html
>
> ------------------------------
>
> Message: 2
> Date: Fri, 30 Jul 2010 20:34:58 +0100
> From: Jing Su <sujing <at> tcd.ie>
> Subject: Re: [Wekalist] help on real value predictions of the number
> of instances
> To: "Weka machine learning workbench list."
> <wekalist <at> list.scms.waikato.ac.nz>
> Message-ID:
> <AANLkTinePPS7BijH+y2kAz7K131T6GK=52fqhcF4fDZC <at> mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
>
> Hello Mark,
>
> Thank you! It is just as you said. If an instance is assigned a positive
> interger N as weight, this instance is actually duplicated N times. This is
> an effective approach to highlight some instances, but this approach easily
> violates original data distribution, is it valid? Whether there are some
> theories about re-weighting to guarantee the validity of result?
>
>
> Thank you very much!
>
>
> Jing
>
> On Wed, Jul 28, 2010 at 9:59 PM, Mark Hall <mhall <at> pentaho.com> wrote:
>
> > On 28/07/10 11:10 AM, Jing Su wrote:
> >
> >> Hello all,
> >>
> >> I prefer to add weight to each instance in ARFF file. Following Weka wiki,
> >> the
> >> new instances are saved in XRFF format. When I use the method below to
> >> load XRFF
> >> file, and run 3-fold cv on naive Bayes classifier, the predicted number of
> >> instances in either class is not integer, as expected. I am confused, and
> >> appreciate any suggestions, Thank you!
> >>
> >
> > If your weights aren't integers, then the confusion matrix will contain
> > non-integer entries as well.
> >
> >
> > Cheers,
> > Mark.
> >
> > _______________________________________________
> > Wekalist mailing list
> > Send posts to: Wekalist <at> list.scms.waikato.ac.nz
> > List info and subscription status:
> > https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
> > List etiquette:
> > http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
> >
> >
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: https://list.scms.waikato.ac.nz/pipermail/wekalist/attachments/20100730/3cf83371/attachment-0001.html
>
> ------------------------------
>
> Message: 3
> Date: Sat, 31 Jul 2010 21:47:21 +1200
> From: Mark Hall <mhall <at> pentaho.com>
> Subject: Re: [Wekalist] one-vs-rest SVM
> To: "Weka machine learning workbench list."
> <wekalist <at> list.scms.waikato.ac.nz>
> Message-ID: <4C53F129.4080803 <at> pentaho.com>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
> icu uci wrote:
> > Hi
> >
> > Does someone have a code about one-vs-rest SVM?
>
> weka.classifiers.meta.MultiClassClassifier implements a number of
> methods to handle multi-class problems with binary class base
> classifiers (e.g. 1-vs-rest, pairwise, error-correcting output codes).
> Source code is available in the Weka distribution.
>
> Cheers,
> Mark.
>
>
>
> ------------------------------
>
> Message: 4
> Date: Sat, 31 Jul 2010 21:50:22 +1200
> From: Mark Hall <mhall <at> pentaho.com>
> Subject: Re: [Wekalist] Function Estimation and Curve Fitting using
> Weka.
> To: "Weka machine learning workbench list."
> <wekalist <at> list.scms.waikato.ac.nz>
> Message-ID: <4C53F1DE.7020904 <at> pentaho.com>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
> Rafik wrote:
> > Hi all,
> >
> > I have a set of m instances ai, i = 1...m, where every instance ai is an
> > n attribute vector (ai1, ai2, ..., aij, ..., ain).
> >
> > I want to estimate an n dimensional function f having a cuve that fits
> > all the m instances ai.
> >
> > After that, I want to find all the instances a'i (from another dataset)
> > that fit the same curve.
> >
> > What is the appropriate way to solve this using Weka ?
>
> You could try model trees - they produce a piecewise linear fit to
> non-linear data. Test points with low error fit the learned model the best.
>
> Cheers,
> Mark.
>
>
>
> ------------------------------
>
> Message: 5
> Date: Sat, 31 Jul 2010 21:52:52 +1200
> From: Mark Hall <mhall <at> pentaho.com>
> Subject: Re: [Wekalist] FW: Arff file format
> To: "Weka machine learning workbench list."
> <wekalist <at> list.scms.waikato.ac.nz>
> Message-ID: <4C53F274.1090209 <at> pentaho.com>
> Content-Type: text/plain; charset=windows-1252; format=flowed
>
> Sarwat Nizamani wrote:
> >
> >
> > Best Regards,
> > *Sarwat Nizamani
> > *
> > PhD Student
> > Maersk Mc-Kinney Moller Institute
> > University of Denmark
> >
> >
> >
> >
> >
> > ------------------------------------------------------------------------
> > From: sarwatnizamani <at> hotmail.com
> > To: wekalist <at> list.scms.waikato.ac.nz
> > Subject: Arff file format
> > Date: Fri, 23 Jul 2010 13:26:14 +0600
> >
> > Dear all, Reply all
> > I am PhD student at Southern University Denmark. I have just started my
> > study and i am new to machine learning my field in PhD is machine
> > learning . I want to use WEKA tool, and I have used it and it is very
> > helpful tool. I want to use WEKA for email classification and WEKA uses
> > ARFF file format for the task. Suppose I have directory that y problms
> > two subdirectories for example one subdirectory named as "Relavant" and
> > other subdirectory named as "Irrrelavant". Each subdirectory contains
> > text files of email.
> > Now I want to convert this directory to ARFF file format as in attached
> > file.
> > Kindly solve my problem I am in a trouble. For the required solution I
> > have tried TextToArffLoader utility, StringToWordVector but Il failed.
> > Please Reply all.
>
> You might need to elaborate a bit on what you mean by "it failed". Was a
> exception thrown or error generated?
>
> Cheers,
> Mark.
>
>
>
> ------------------------------
>
> Message: 6
> Date: Sat, 31 Jul 2010 15:44:47 +0530
> From: shehan <at> uom.lk
> Subject: [Wekalist] Problem in uploading Supplied Test Set in ver
> 3.6.3
> To: wekalist <at> list.scms.waikato.ac.nz
> Message-ID: <20100731154447.20305qum6dgz3vy8 <at> webmail.uom.lk>
> Content-Type: text/plain; charset=ISO-8859-1; DelSp="Yes";
> format="flowed"
>
>
> There seems to be a problem in uploading "Supplied test set" on Weka
> 3.6.3 in the classification module. It is not detecting the data
> instances from the input file. It is reporting 0 instances and the
> correct number of attributes. Tried both arff and csv input files and
> also the linux and windows versions.
>
> It works on version 3.4.
>
> thx
>
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
>
>
>
>
>
> ------------------------------
>
> Message: 7
> Date: Sat, 31 Jul 2010 14:59:47 +0330
> From: m lalshateri <ma.lalshateri <at> gmail.com>
> Subject: [Wekalist] extension of weka
> To: wekalist <wekalist <at> list.scms.waikato.ac.nz>
> Message-ID:
> <AANLkTimuZZORaA6sVEU7VZ0iB38i9Wvph5Bj03yCHE7f <at> mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> Hello
> I use NetBeans IDE6.5 and i have 2 questions:
>
> 1. I wrote an algorithm and now i decided to add it to the weka(that
> users be able to select it from explorer user interface). How can i do
> it?
>
> 2. to reach above goal i have to create a ".jar" file (exactly
> "weka.jar"). So i added a code to "build.xml" and did some thing else.
> But after compile and build project following message is shown:
>
> "The system is out of resources.
> Consult the following stack trace for details.
> java.lang.OutOfMemoryError: Java heap space
> BUILD FAILED"
>
> what should i do?!
>
> thanks in advance
>
>
>
> ------------------------------
>
> Message: 8
> Date: Sat, 31 Jul 2010 23:43:47 +0800
> From: icu uci <icushj <at> gmail.com>
> Subject: [Wekalist] what is the unit of time used in weka.experimenter
> To: wekalist <at> list.scms.waikato.ac.nz
> Message-ID:
> <AANLkTikREFZOhKu_yJgopiJfZPm4=7cr7LriAgxi+yxM <at> mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
>
> I use the experimenter to analyze the running result, and I choose
> userCPUtraining time option to compare the results. I want to know what is
> the unit of the time here? Thx.
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: https://list.scms.waikato.ac.nz/pipermail/wekalist/attachments/20100731/8099c0c4/attachment-0001.html
>
> ------------------------------
>
> Message: 9
> Date: Sat, 31 Jul 2010 22:57:01 +0530
> From: Ruchi Rana <ruchirana86 <at> gmail.com>
> Subject: Re: [Wekalist] extension of weka
> To: "Weka machine learning workbench list."
> <wekalist <at> list.scms.waikato.ac.nz>
> Message-ID:
> <AANLkTi=RkODC9d90P_phYWa6inSnZfmWhjU+znGqb=Wz <at> mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
>
> Hi
> you can refer the following link, may be you will get the
> solution...
>
> http://weka.wikispaces.com/Netbeans+6.0+%28weka-src.jar%29
>
>
>
>
> On Sat, Jul 31, 2010 at 4:59 PM, m lalshateri <ma.lalshateri <at> gmail.com>wrote:
>
> > Hello
> > I use NetBeans IDE6.5 and i have 2 questions:
> >
> > 1. I wrote an algorithm and now i decided to add it to the weka(that
> > users be able to select it from explorer user interface). How can i do
> > it?
> >
> > 2. to reach above goal i have to create a ".jar" file (exactly
> > "weka.jar"). So i added a code to "build.xml" and did some thing else.
> > But after compile and build project following message is shown:
> >
> > "The system is out of resources.
> > Consult the following stack trace for details.
> > java.lang.OutOfMemoryError: Java heap space
> > BUILD FAILED"
> >
> > what should i do?!
> >
> > thanks in advance
> >
> > _______________________________________________
> > Wekalist mailing list
> > Send posts to: Wekalist <at> list.scms.waikato.ac.nz
> > List info and subscription status:
> > https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
> > List etiquette:
> > http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html<http://www.cs.waikato.ac.nz/%7Eml/weka/mailinglist_etiquette.html>
> >
>
>
>
> --
> Regards
> Ruchi Rana
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: https://list.scms.waikato.ac.nz/pipermail/wekalist/attachments/20100731/31e53239/attachment.html
>
> ------------------------------
>
> _______________________________________________
> Wekalist mailing list
> Wekalist <at> list.scms.waikato.ac.nz
> https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
>
>
> End of Wekalist Digest, Vol 90, Issue 1
> ***************************************
_______________________________________________
Wekalist mailing list
Send posts to: Wekalist <at> list.scms.waikato.ac.nz
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Permalink | Reply | 0 comments |
headers
Bonnieh | 2 Aug 2010 03:37
Picon
Picon
Favicon

can't find DatabaseUtils.props in weka installation directory


Hi,

I have installed weka 3.6.3 in Windows Vista. I have set the build path to
eclipse, and class path for mysql driver and weka.jar. But I can't find
DatabaseUtils.props to change the drvier and URL. I have search the whole PC
and couldn't find the file. It can be viewed from eclipse but can't be
changed from there. Can you please help?
--

-- 
View this message in context: http://old.nabble.com/MySQL-and-Weka-tp6600635p29321488.html
Sent from the WEKA mailing list archive at Nabble.com.


_______________________________________________
Wekalist mailing list
Send posts to: Wekalist <at> list.scms.waikato.ac.nz
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Permalink | Reply | 0 comments |
headers
Bonnieh | 2 Aug 2010 03:41
Picon
Picon
Favicon

can't find DatabaseUtils.props


Hi,

I have installed weka 3.6.3 in Windows Vista. I have set the build path to
eclipse, and class path for mysql driver and weka.jar. But I can't find
DatabaseUtils.props to change the drvier and URL. I have search the whole PC
and couldn't find the file. It can be viewed from eclipse but can't be
changed from there. Can someone please help?

--

-- 
View this message in context: http://old.nabble.com/can%27t-find-DatabaseUtils.props-tp29321500p29321500.html
Sent from the WEKA mailing list archive at Nabble.com.


_______________________________________________
Wekalist mailing list
Send posts to: Wekalist <at> list.scms.waikato.ac.nz
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Permalink | Reply | 6 comments |
headers
tabuchi, takuzou - MPS | 2 Aug 2010 03:56
Picon
Favicon

RE: can't find DatabaseUtils.props

Hi,
DatabaseUtils.props is in the weka.jar
under  weka/experiment/

-----Original Message-----
From: wekalist-bounces <at> list.scms.waikato.ac.nz
[mailto:wekalist-bounces <at> list.scms.waikato.ac.nz] On Behalf Of Bonnieh
Sent: Monday, August 02, 2010 10:41 AM
To: wekalist <at> list.scms.waikato.ac.nz
Subject: [Wekalist] can't find DatabaseUtils.props

Hi,

I have installed weka 3.6.3 in Windows Vista. I have set the build path
to
eclipse, and class path for mysql driver and weka.jar. But I can't find
DatabaseUtils.props to change the drvier and URL. I have search the
whole PC
and couldn't find the file. It can be viewed from eclipse but can't be
changed from there. Can someone please help?

--

-- 
View this message in context:
http://old.nabble.com/can%27t-find-DatabaseUtils.props-tp29321500p293215
00.html
Sent from the WEKA mailing list archive at Nabble.com.


_______________________________________________
Wekalist mailing list
Send posts to: Wekalist <at> list.scms.waikato.ac.nz
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Permalink | Reply | 5 comments |
headers
Wen (Bonnie) Hoschke | 2 Aug 2010 04:01
Picon
Picon
Favicon

RE: can't find DatabaseUtils.props

Thanks for your reply, I can see weka.jar, just I can't extract it. It
said missing manifest for main class. 

-----Original Message-----
From: wekalist-bounces <at> list.scms.waikato.ac.nz
[mailto:wekalist-bounces <at> list.scms.waikato.ac.nz] On Behalf Of tabuchi,
takuzou - MPS
Sent: Monday, August 02, 2010 11:57 AM
To: Weka machine learning workbench list.
Subject: RE: [Wekalist] can't find DatabaseUtils.props

Hi,
DatabaseUtils.props is in the weka.jar
under  weka/experiment/

-----Original Message-----
From: wekalist-bounces <at> list.scms.waikato.ac.nz
[mailto:wekalist-bounces <at> list.scms.waikato.ac.nz] On Behalf Of Bonnieh
Sent: Monday, August 02, 2010 10:41 AM
To: wekalist <at> list.scms.waikato.ac.nz
Subject: [Wekalist] can't find DatabaseUtils.props

Hi,

I have installed weka 3.6.3 in Windows Vista. I have set the build path
to eclipse, and class path for mysql driver and weka.jar. But I can't
find DatabaseUtils.props to change the drvier and URL. I have search the
whole PC and couldn't find the file. It can be viewed from eclipse but
can't be changed from there. Can someone please help?

--
View this message in context:
http://old.nabble.com/can%27t-find-DatabaseUtils.props-tp29321500p293215
00.html
Sent from the WEKA mailing list archive at Nabble.com.


_______________________________________________
Wekalist mailing list
Send posts to: Wekalist <at> list.scms.waikato.ac.nz
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Permalink | Reply | 4 comments |
headers
nevre | 2 Aug 2010 07:40
Picon

RE: can't find DatabaseUtils.props

Try opening it with a compression program like winzip or 7-zip or winRAR.
If you double click on it to open it you will be trying to execute it.

If you can't open it with one of those programs your file may be corrupt.

> Thanks for your reply, I can see weka.jar, just I can't extract it. It
> said missing manifest for main class.
>
> -----Original Message-----
> From: wekalist-bounces <at> list.scms.waikato.ac.nz
> [mailto:wekalist-bounces <at> list.scms.waikato.ac.nz] On Behalf Of tabuchi,
> takuzou - MPS
> Sent: Monday, August 02, 2010 11:57 AM
> To: Weka machine learning workbench list.
> Subject: RE: [Wekalist] can't find DatabaseUtils.props
>
> Hi,
> DatabaseUtils.props is in the weka.jar
> under  weka/experiment/
>
> -----Original Message-----
> From: wekalist-bounces <at> list.scms.waikato.ac.nz
> [mailto:wekalist-bounces <at> list.scms.waikato.ac.nz] On Behalf Of Bonnieh
> Sent: Monday, August 02, 2010 10:41 AM
> To: wekalist <at> list.scms.waikato.ac.nz
> Subject: [Wekalist] can't find DatabaseUtils.props
>
>
> Hi,
>
> I have installed weka 3.6.3 in Windows Vista. I have set the build path
> to eclipse, and class path for mysql driver and weka.jar. But I can't
> find DatabaseUtils.props to change the drvier and URL. I have search the
> whole PC and couldn't find the file. It can be viewed from eclipse but
> can't be changed from there. Can someone please help?
>
> --
> View this message in context:
> http://old.nabble.com/can%27t-find-DatabaseUtils.props-tp29321500p293215
> 00.html
> Sent from the WEKA mailing list archive at Nabble.com.
>
>
>
>
> _______________________________________________
> Wekalist mailing list
> Send posts to: Wekalist <at> list.scms.waikato.ac.nz
> List info and subscription status:
> https://list.scms.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: Wekalist <at> list.scms.waikato.ac.nz
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Permalink | Reply | 3 comments |
headers
Wen (Bonnie) Hoschke | 2 Aug 2010 09:13
Picon
Picon
Favicon

RE: can't find DatabaseUtils.props

Hi,

I have extracted the weka.jar with jar.exe. I have changed the
DatabaseUtils.props accordingly. I still got the error message below:

 Trying to add database driver (JDBC): RmiJdbc.RJDriver - Error, not in
CLASSPATH?

Trying to add database driver (JDBC): jdbc.idbDriver - Error, not in
CLASSPATH?

Trying to add database driver (JDBC): com.mckoi.JDBCDriver - Error, not
in CLASSPATH?

Trying to add database driver (JDBC): org.hsqldb.jdbcDriver - Error, not
in CLASSPATH?

java.util.Vector$1 <at> 12ac982

Exception in thread "main" java.sql.SQLException: No suitable driver
found for jdbc:idb=experiments.prp

      at java.sql.DriverManager.getConnection(Unknown Source)

      at java.sql.DriverManager.getConnection(Unknown Source)

      at
weka.experiment.DatabaseUtils.connectToDatabase(DatabaseUtils.java:521)

      at
weka.experiment.InstanceQuery.retrieveInstances(InstanceQuery.java:286)

      at
weka.experiment.InstanceQuery.retrieveInstances(InstanceQuery.java:271)

      at load.main(load.java:17)

I am not sure where is wronf, can you please help? Thanks heaps

Cheers
Bonnie

-----Original Message-----
From: wekalist-bounces <at> list.scms.waikato.ac.nz
[mailto:wekalist-bounces <at> list.scms.waikato.ac.nz] On Behalf Of
nevre <at> dons.net.au
Sent: Monday, August 02, 2010 3:40 PM
To: Weka machine learning workbench list.
Subject: RE: [Wekalist] can't find DatabaseUtils.props

Try opening it with a compression program like winzip or 7-zip or
winRAR.
If you double click on it to open it you will be trying to execute it.

If you can't open it with one of those programs your file may be
corrupt.

> Thanks for your reply, I can see weka.jar, just I can't extract it. It

> said missing manifest for main class.
>
> -----Original Message-----
> From: wekalist-bounces <at> list.scms.waikato.ac.nz
> [mailto:wekalist-bounces <at> list.scms.waikato.ac.nz] On Behalf Of 
> tabuchi, takuzou - MPS
> Sent: Monday, August 02, 2010 11:57 AM
> To: Weka machine learning workbench list.
> Subject: RE: [Wekalist] can't find DatabaseUtils.props
>
> Hi,
> DatabaseUtils.props is in the weka.jar under  weka/experiment/
>
> -----Original Message-----
> From: wekalist-bounces <at> list.scms.waikato.ac.nz
> [mailto:wekalist-bounces <at> list.scms.waikato.ac.nz] On Behalf Of Bonnieh
> Sent: Monday, August 02, 2010 10:41 AM
> To: wekalist <at> list.scms.waikato.ac.nz
> Subject: [Wekalist] can't find DatabaseUtils.props
>
>
> Hi,
>
> I have installed weka 3.6.3 in Windows Vista. I have set the build 
> path to eclipse, and class path for mysql driver and weka.jar. But I 
> can't find DatabaseUtils.props to change the drvier and URL. I have 
> search the whole PC and couldn't find the file. It can be viewed from 
> eclipse but can't be changed from there. Can someone please help?
>
> --
> View this message in context:
> http://old.nabble.com/can%27t-find-DatabaseUtils.props-tp29321500p2932
> 15
> 00.html
> Sent from the WEKA mailing list archive at Nabble.com.
>
>
>
>
> _______________________________________________
> Wekalist mailing list
> Send posts to: Wekalist <at> list.scms.waikato.ac.nz List info and 
> subscription status:
> https://list.scms.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: Wekalist <at> list.scms.waikato.ac.nz List info and
subscription status:
https://list.scms.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: Wekalist <at> list.scms.waikato.ac.nz
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Permalink | Reply | 2 comments |
headers
nevre | 2 Aug 2010 09:40
Picon

RE: can't find DatabaseUtils.props

> I have extracted the weka.jar with jar.exe. I have changed the
> DatabaseUtils.props accordingly. I still got the error message below:
>
>  Trying to add database driver (JDBC): RmiJdbc.RJDriver - Error, not in
> CLASSPATH?
>
> Trying to add database driver (JDBC): jdbc.idbDriver - Error, not in
> CLASSPATH?
>
> Trying to add database driver (JDBC): com.mckoi.JDBCDriver - Error, not
> in CLASSPATH?
>
> Trying to add database driver (JDBC): org.hsqldb.jdbcDriver - Error, not
> in CLASSPATH?

These are just warnings and only apply if you are using database drivers
for one of this particular types of database. Warning messages, not error
messages.

> Exception in thread "main" java.sql.SQLException: No suitable driver
> found for jdbc:idb=experiments.prp
>       at java.sql.DriverManager.getConnection(Unknown Source)
>       at java.sql.DriverManager.getConnection(Unknown Source)
>       at
> weka.experiment.DatabaseUtils.connectToDatabase(DatabaseUtils.java:521)
>       at
> weka.experiment.InstanceQuery.retrieveInstances(InstanceQuery.java:286)
>       at
> weka.experiment.InstanceQuery.retrieveInstances(InstanceQuery.java:271)
>       at load.main(load.java:17)
>
> I am not sure where is wronf, can you please help? Thanks heaps
>

It looks like you are trying to use experimenter. It would probably help
to explain a bit about what you are trying to achieve as well as posting
the things that go wrong, otherwise we are really just guessing.

Did you modify the databaseutils.props file? What kind of database are you
trying to connect to? The line above
jdbc:idb=experiments.prp
seems to suggest that you left the default settings alone in experimenter?
You will have to change that line in the experimenter to whatever database
your are trying to connect to.....

For example, I am using Microsoft SQL Server.

I altered the line in the databaseutils.props file to read
# database URL
jdbcURL=jdbc:odbc:wekabase

Then when I use experimenter I selected JDBC database as the results
destination and put jdbc:odbc:wekabase in as the url. I click on the
browse button next to the url and enter my username and password.

Before this I set up wekabase as and ODBC database in my windows
installation. There is a webpage in the wiki about doing this.

Hope you can get it working!

_______________________________________________
Wekalist mailing list
Send posts to: Wekalist <at> list.scms.waikato.ac.nz
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Permalink | Reply | 1 comment |
headers
Mark Hall | 2 Aug 2010 11:38
Favicon

Re: SMO quadratic SVM output

ninjasmith wrote:
> 
> 
> dqnaiman wrote:
>> Dear all,
>>
>> Can someone please tell me how to interpret the output of the Weka SMO
>> classifier with exponent 2?
>> I have built a simple binary classifier using two predictive features,
>> call them x1 and x2.
>> I'm looking for an explanation that says how to interpret this output:
>>
>>      0.457  * <0 1 > * X]
>>  +       0.543  * <1 0.761104 > * X]
>>  -       1      * <0.920564 0 > * X]
>>  +       0.2284
>>
>> that reads something like this 
>>
>> "To classify (x1,x2) you calculate the following quantity ____ and see if 
>> it is positive or not."

The output is a weighed combination of kernel evaluations involving the 
support vectors (shown in angle brackets) and a test point X. For a test 
point and the first support vector you compute the dot product between 
the two and then apply the kernel function to the result (in this case, 
for a polynomial kernel with exponent two, you square the result). Then 
you multiply this by the weight (0.457 for the first support vector). 
Then repeat this process for the other support vectors and add them all 
together and add on the bias term.

Cheers,
Mark.

_______________________________________________
Wekalist mailing list
Send posts to: Wekalist <at> list.scms.waikato.ac.nz
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Permalink | Reply | 0 comments |
headers
Mark Hall | 2 Aug 2010 11:45
Favicon

Re: interpretting the support vectors, the kernel and the logistic regression coefficients

ninjasmith wrote:
> Hi,
> 
> I understand how SVMs work.  in that the kernel transforms a linear
> hyperplane into a curved hyperplane in featurespace.
> 
> What I don't understand is how to interpret the output of weka explorer.
> 
> for instance is I am using poly kernel with exp 2 my kernel is
> 
> K(x,y) = <x,y>^2.0
> 
> and I have a bunch of support vectors
> 
> 1 * <0.001743 00834 0.06881 > * X ]
> + 0.4371 * <...  .... ...> * X]
> + etc...
> 
> so given a new unknown instance how do I classify.  presumably I multiply by
> each support vector and apply the kernel formula at some point and sum and
> then if its positive or negative I know the class.

See my answer to an earlier posting on this.

> 
> also how does the Logistic regression work.  I am expecting it to give me a
> psuedo probablility for the class. i.e. a measure of the distance of the
> point from the hyperplane.  I'm not sure how to get this from information
> like the following (where snare is one of my two classes)

Logistic regression is simply applied to a dataset that consists of the 
output of the SVM as an attribute and the class label.

Cheers,
Mark.

_______________________________________________
Wekalist mailing list
Send posts to: Wekalist <at> list.scms.waikato.ac.nz
List info and subscription status: https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Permalink | Reply | 1 comment |

Gmane