Re: Weka-Spark

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

Re: Weka-Spark

MOHAMMED KAMAL

Hi Eibe, 

> when i put sampling techniques only in weka spark job and remove it from weka spark evaluation job (no change in values ) .. why ?

Q1. Can you be more specific about what you did?

__________________________________________


i put  the following sequence : 
arff header spark job (4 data slices)  ---> randomly shuffle data spark job  -----> i add  sampling technique (SMOTE for example) + weka classifier spark job --------->   weka classifier evaluation without sampling (as you recommensd me before)

nothing change in accuracy and f-measure ?? 
if i put sampling also in weka classification evaluation spark job i get higher accuracy and f-measure 
note: you recommend me to do not that

________________________________________

Q2. if i compare 10 classifiers without sampling and with sampling techniques (class balancer-ROS-RUS-Smote)  on imbalanced dataset what measure should i put in the table for comparison ?

is it enough to put accuracy , average f-measure and f measure for minority class ? what kind of graphics will be demonstrative for academic papers ? 

------------------------------------------------------------------------------

Q3, i asked about  ROC curve to compare these classifiers 
your reply : The Explorer uses the same code as the KnowledgeFlow. Yes, you can use the KnowledgeFlow to plot ROC curves. You can also save the data for a plot into a file, using the Save? button in the plotting window. Then you can use an external tool to make a high-resolution plot.
please can you give me further information, i didn't understand this point
do you mean to make a plot for every classifier and save it separately and then aggregate all with another tool?   which tool do you recommend ?   how can i do this?  

i have 10 classifiers ( without - with CB-RUS-ROS-SMOTE)
50 readings , so will be huge to compare figures 
i think i can give a figure compare each classifier with sampling with / without sampling techniques using ROC curve? 
or do you recommend to make bar chart for F-Measure in minority class only
i want your recommendation 


cheers 

M.kamal









From: [hidden email] <[hidden email]> on behalf of [hidden email] <[hidden email]>
Sent: Wednesday, June 7, 2017 11:23 AM
To: [hidden email]
Subject: Wekalist Digest, Vol 172, Issue 24
 
Send Wekalist mailing list submissions to
        [hidden email]

To subscribe or unsubscribe via the World Wide Web, visit
        https://list.waikato.ac.nz/mailman/listinfo/wekalist


or, via email, send a message with subject or body 'help' to
        [hidden email]

You can reach the person managing the list at
        [hidden email]

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Wekalist digest..."


Today's Topics:

   1. Re: LSA implementation (Andria Lan)
   2. Re: Weka-Spark (Eibe Frank)
   3. Re: AutoWEKA on Windows (Eibe Frank)


----------------------------------------------------------------------

Message: 1
Date: Wed, 7 Jun 2017 09:36:59 +0800
From: Andria Lan <[hidden email]>
To: "Weka machine learning workbench list."
        <[hidden email]>
Subject: Re: [Wekalist] LSA implementation
Message-ID:
        <CAHM6HB0gqV2YGySomFSd0ToMObVm0Vg7KXDWEaM-W7vhekS=[hidden email]>
Content-Type: text/plain; charset="utf-8"

Many thanks, Eibe. Highly appreciat your extensive help.

Good luck to you.

Kind regards,
Andria
On 6 Jun 2017 6:48 a.m., "Eibe Frank" <[hidden email]> wrote:

>
> > On 5/06/2017, at 10:27 PM, Andria Lan <[hidden email]> wrote:
> >
> > Thanks a lot for the extensive information. Finally, just wanted to
> confirm the following issue:
> >
> > > GridSearch only works for exactly two parameters, and
> CVParameterSelection cannot optimise nested parameters.
> >
> > Did you mean that MultiSearch works similar to GridSearch for the goal
> of parameters optimisation. Right?
>
> Yes. GridSearch has the option to extend the grid if necessary, which
> MultiSearch doesn?t have, but it can only deal with optimisation problems
> involving exactly two parameters.
>
> Cheers,
> Eibe
>
>
> _______________________________________________
> Wekalist mailing list
> Send posts to: [hidden email]
> List info and subscription status: https://list.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: <http://list.waikato.ac.nz/pipermail/wekalist/attachments/20170607/0a9417a3/attachment-0001.html>

------------------------------

Message: 2
Date: Wed, 7 Jun 2017 16:47:54 +1200
From: Eibe Frank <[hidden email]>
To: "Weka machine learning workbench list."
        <[hidden email]>
Subject: Re: [Wekalist] Weka-Spark
Message-ID: <[hidden email]>
Content-Type: text/plain; charset=utf-8


> On 7/06/2017, at 8:34 AM, ENGMohammed kamal <[hidden email]> wrote:
>
> when i put sampling techniques only in weka spark job and remove it from weka spark evaluation job (no change in values ) .. why ?

Can you be more specific about what you did?

> i have more 3 questions
> 1- if i have 10 classifiers and iwant to draw ROC curve for them for comparison ( with best high resolution )  , you told me befor this is not available in knowledge flow , how can i do this using explorer or experminter ?  note : i try normal weka o/p  it's not clear and most of line overlapped . with sympols like + & *

The Explorer uses the same code as the KnowledgeFlow. Yes, you can use the KnowledgeFlow to plot ROC curves. You can also save the data for a plot into a file, using the Save? button in the plotting window. Then you can use an external tool to make a high-resolution plot.

> 2-  i worked with vote classifiers vote ensemble in weka spark  , the output model for JRip ( contains 4 model) as i sliced my dataset to 4 partitions ( please see attached file) , but how can i know the best winning model between the 4 models by voting.

Voting will use all the models. It?s does not use a single model for prediction.

> 3-  if i reference my work in research paper for experiments which are done by weka -spark, it is enough these two references or i need more , specially for weka-spark ??
> M.Hall, Mark, Eibe Frank, Geoffrey Holmes, Bernhard Pfahringer, Peter Reutemann, and Ian H. Witten. "The WEKA data mining software: an update." ACM SIGKDD explorations newsletter 11, no. 1 (2009): 10-18.
>
> [69] Eibe Frank, Mark A. Hall, and Ian H. Witten (2016). The WEKA Workbench. Online Appendix for "Data Mining: Practical Machine Learning Tools and Techniques", Morgan Kaufmann, Fourth Edition, 2016.

That sounds fine.

Cheers,
Eibe



------------------------------

Message: 3
Date: Wed, 7 Jun 2017 20:22:59 +1200
From: Eibe Frank <[hidden email]>
To: "Weka machine learning workbench list."
        <[hidden email]>
Subject: Re: [Wekalist] AutoWEKA on Windows
Message-ID: <[hidden email]>
Content-Type: text/plain; charset=utf-8

I use WEKA on Windows 10 all the time, and Windows is the most popular platform for running WEKA. There must be something specific to your particular set-up that's causing problems.

Cheers,
Eibe

> On 6 Jun 2017, at 21:32, Alexander Osherenko <[hidden email]> wrote:
>
> The problem in Weka installation on Win10 is RunWeka.bat. If I comment out the first line in it I can see that last line cant't be executed if called on the console from outside the Weka directory (works good if called from the Weka directory):
>
> javaw -classpath . RunWeka -i .\RunWeka.ini -w .\weka.jar -c default ""
>
> Consequently, although the Weka shortcut specifies the Run in directory, the target ..\Weka-3-8\RunWeka.bat default doesn't work. Specifying instead the target ..\Weka-3-8\debug_runWeka.bat (my own) that contains one line
>
> java -jar ...\Weka-3-8\weka.jar
>
> shows up Weka started from Win10 menu.
>
> Best, Alexander?
>
> 2017-06-05 12:43 GMT+01:00 Alexander Osherenko <[hidden email]>:
> There are problems with paths on Win10.
>
> 1. It begins with installing Weka that can't start from menu although the shortcut seems to be OK: Target and Start in folder are correct, but RunWeka.bat doesn't work. Windows console flashes but doesn't start Weka -- I assume, Windows doesn't find RunWeka.bat or RunWeka.bat crashes.
>
> ?Whatsoever, ??I developed a batch ??workaround ?file that does exactly the same as the shortcut should:
> set _JAVA_OPTIONS=-Djava.class.path=./weka.jar;
> cd D:\toDeleteJDK\Weka-3-8
> d:
> D:\toDeleteJDK\Weka-3-8\RunWeka.bat default
>
> 2. This starts Weka and EVEN the Weka Explorer )). But there are other problems: The AutoWeka package is installed on the hard drive but not in Weka -- AutoWeka classes are not found what I can see on the Weka log, for example, instantiation of AutoWEKA classes fails. I added autoweka.jar manually to the first line of my batch file.
>
> _JAVA_OPTIONS=-Djava.class.path=./weka.jar;C:\Users\osher\wekafiles\packages\Auto-WEKA\autoweka.jar;
>
> and now Weka finds necessary AutoWeka classes but the path problem is not solved: I see no instantiation errors on the log anymore, but other error messages (AutoWeka complains about failures), for example,
> Experiment - Failed to parse trajectory
>
> There are no errors on XP at all (neither with the Weka shortcut, nor with the missing autoweka.jar). I assume there are problems with incorrect settings in Win10, but the Weka and the AutoWeka reinstalls to update paths don't help.
>
> ?Best, Alexander
> ?
> 2017-06-05 5:30 GMT+01:00 Eibe Frank <[hidden email]>:
> Yes, on my Mac, the following works:
>
> export _JAVA_OPTIONS="-Dsun.java.command=dummy -Djava.home=/Library/Java/JavaVirtualMachines/jdk1.8.0_102.jdk/Contents/Home/jre"
>
> Cheers,
> Eibe
>
> PS: Interestingly, I had to set -Djava.home to .../jre to make WEKA work, but the /jre part can be omitted when using the JAVA_HOME environment variable instead.
>
>
> > On 5 Jun 2017, at 07:08, Peter Reutemann <[hidden email]> wrote:
> >
> > On June 5, 2017 2:57:55 AM GMT+12:00, Alexander Osherenko <[hidden email]> wrote:
> > Is it possible to set several variables in JAVA_OPTIONS, for example,
> > to set _JAVA_OPTIONS to sun.java.command=var1 and java.home=var2?
> >
> > If I specify string -Dsun.java.command=var1;java.home=var2, it is not parsed correctly in Systeminfo.
> >
> > Cheers, Alexander
> >
> > 2017-06-03 8:59 GMT+01:00 Alexander Osherenko <[hidden email]>:
> > Got it, thanx. It is also possible to set the sun.java.command in Systeminfo by setting
> > _JAVA_OPTIONS=-Dsun.java.command="dummy".
> >
> > Best, Alexander
> >
> > 2017-06-03 5:44 GM??T+01:00 Eibe Frank <[hidden email]>:
> >
> > Yes, modifying RunWeka.bat should work too.
> >
> > The advantage of using an environment variable is that you won't have to fix this problem again when you install a different version of WEKA. Also, it works even if you don't have admin rights. A correct setting for JAVA_HOME may also be useful for other applications, so from that point of view it may also make sense to set this globally in an environment variable.
> >
> > The advantage of using _JAVA_OPTIONS in particular is that any settings specified in this variable take precedence over other settings provided for the java command, see
> >
> > https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts


> >
> > Cheers,
> > Eibe
> >
> > > On 3 Jun 2017, at 16:08, Alexander Osherenko <[hidden email]> wrote:
> > >
> > > I wonder why you are talking about this variable. Is it the only way to set java.home variable in Weka or I can set it in RunWeka.bat?
> > >
> > > ?Cheers, Alexander
> > > ?
> > > 2017-06-02 23:10 GMT+01:00 Eibe Frank <[hidden email]>:
> > > Try setting a new variable
> > >
> > >   _JAVA_OPTIONS
> > >
> > > as a Windows environment variable and give it the value
> > >
> > >   -Djava.home=...
> > >
> > > Cheers,
> > > Eibe
> > >
> > >
> > >
> > > Sent from my iPad
> > > On 3/06/2017, at 7:15 AM, Alexander Osherenko <[hidden email]> wrote:
> > >
> > >> I tried to find the problem and didn't. Maybe you remember I have Weka 3-8-1 and AutoWEKA installed and they are running without any problems on XP. That's why I have to seek further.
> > >>
> > >> Let us please compare the settings in Weka Systeminfo. I found some differences, for example, in java.home and sun.java.command. Can I set these variables in RunWeka.bat?
> > >>
> > >> 2017-06-02 9:24 GMT+01:00 Michael Hall <[hidden email]>:
> > >> > On Jun 1, 2017, at 8:48 PM, Eibe Frank <[hidden email]> wrote:
> > >> >
> > >> > With Oracle JVMs, you can use the _JAVA_OPTIONS environment variable to pass such command-line parameters to the JVM. This works on all platforms.
> > >>
> > >>
> > >> I wasn?t aware of that one. It should work for this then, I would think. Unless Java itself sets java.home and overrides the one you specify.
> > >>
> > >> I didn?t entirely follow this thread and I am not sure why it is necessary. A jvm to run the application is obviously being found and if the same one is used to set the java.home property you would think AutoWeka would work with that? Possibly some localization issue as the original poster indicated. Then setting the property could work.
> > >>
> > >> The other possibility would be that it is the same situation as on OS X. For that the ?java? executable command is not included in the Weka application embedded jre. So AutoWeka can?t find it there and fails to Runtime exec java. I still believe this is the behavior of the appbundler version that is being used to create the application bundle. Probably just to save some space the executables are omitted. I also still think a fix there would be for whatever is used for the application build process to create a ?bin' directory in the embedded jre and copy in the ?java? executable command. This did in fact work for me, doing it myself, to get AutoWeka going on OS X.
> > >>
> > >> For what this has to do with Windows, if you try to override java.home I would make sure you are pointing it at a jre/jdk that includes bin/java or bin/java.exe.
> > >>
> > >> Michael Hall
> > >>
> > >>
> > >> _______________________________________________
> > >> Wekalist mailing list
> > >> Send posts to: [hidden email]
> > >> List info and subscription status: 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]
> > >> List info and subscription status: 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]
> > > List info and subscription status: 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]
> > > List info and subscription status: 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]
> > List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist


> > List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html


> >
> >
> >
> > Separate the options with blanks instead of semicolons. This variable acts like a partial command-line string, which gets added to the actual command-line to start the jvm.
> >
> > Cheers, Peter
> > --
> > Peter Reutemann
> > Dept. of Computer Science
> > University of Waikato, NZ
> > +64 (7) 858-5174
> > http://www.cms.waikato.ac.nz/~fracpete/


> > http://www.data-mining.co.nz


> > _______________________________________________
> > Wekalist mailing list
> > Send posts to: [hidden email]
> > List info and subscription status: 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]
> List info and subscription status: 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]
> List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist


> List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html





------------------------------

_______________________________________________
Wekalist mailing list
[hidden email]
https://list.waikato.ac.nz/mailman/listinfo/wekalist




End of Wekalist Digest, Vol 172, Issue 24
*****************************************

_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: 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: Weka-Spark

Mark Hall
I don't think there is a problem using SMOTE in the evaluation job, as only the partial training folds will be affected by the SMOTE process. What is the purpose of running the classifier job before the evaluation job? Is this just to create a final model?

Cheers,
Mark.


On 19/06/17, 9:18 AM, "ENGMohammed kamal" <[hidden email] on behalf of [hidden email]> wrote:

    Hi Eibe,
    >
    > when i put
     sampling techniques only in weka spark job and remove
     it from weka spark evaluation job (no change in values
     ) .. why ?
   
    Q1. Can you be more specific about what you did?
    __________________________________________
   
   
    i put  the following sequence :
    arff header spark job (4 data slices)  ---> randomly shuffle data spark job  ----->
     i add  sampling technique (SMOTE for example) + weka classifier spark job --------->   weka classifier evaluation without sampling (as you recommensd me before)
   
    nothing change in accuracy and f-measure ??
    if i put sampling also in weka classification evaluation spark job i get higher accuracy and f-measure
    note: you recommend me to do not that
    ________________________________________
    Q2. if i compare 10 classifiers without sampling and with sampling techniques (class balancer-ROS-RUS-Smote)  on imbalanced dataset what measure should i put in the table for comparison ?
    is it enough to put accuracy , average f-measure and f measure for minority class ? what kind of graphics will be demonstrative for academic papers ?
    ------------------------------------------------------------------------------
    Q3, i asked about  ROC curve to compare these classifiers
    your reply : The Explorer uses the same code as the KnowledgeFlow.
     Yes, you can use the KnowledgeFlow to plot ROC curves. You can also save the data for a plot into a file, using the Save? button in the plotting window. Then you can use an external tool to make a high-resolution plot.
    please can you give me further information, i didn't understand this point
    do you mean to make a plot for every classifier and save it separately and then aggregate all with another tool?   which tool do you recommend ?   how can i do this?  
    i have 10 classifiers ( without - with CB-RUS-ROS-SMOTE)
    50 readings , so will be huge to compare figures
    i think i can give a figure compare each classifier with sampling with / without sampling techniques using ROC curve?
    or do you recommend to make bar chart for F-Measure in minority class only
    i want your recommendation
   
   
    cheers
    M.kamal
   
   
   
   
   
   
   
   
   
   
    ________________________________________
    From: [hidden email] <[hidden email]> on behalf of [hidden email] <[hidden email]>
    Sent: Wednesday, June 7, 2017 11:23 AM
    To: [hidden email]
    Subject: Wekalist Digest, Vol 172, Issue 24
   
   
    Send Wekalist mailing list submissions to
            [hidden email]
   
    To subscribe or unsubscribe via the World Wide Web, visit
           
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    or, via email, send a message with subject or body 'help' to
            [hidden email]
   
    You can reach the person managing the list at
            [hidden email]
   
    When replying, please edit your Subject line so it is more specific
    than "Re: Contents of Wekalist digest..."
   
   
    Today's Topics:
   
       1. Re: LSA implementation (Andria Lan)
       2. Re: Weka-Spark (Eibe Frank)
       3. Re: AutoWEKA on Windows (Eibe Frank)
   
   
    ----------------------------------------------------------------------
   
    Message: 1
    Date: Wed, 7 Jun 2017 09:36:59 +0800
    From: Andria Lan <[hidden email]>
    To: "Weka machine learning workbench list."
            <[hidden email]>
    Subject: Re: [Wekalist] LSA implementation
    Message-ID:
            <CAHM6HB0gqV2YGySomFSd0ToMObVm0Vg7KXDWEaM-W7vhekS=[hidden email]>
    Content-Type: text/plain; charset="utf-8"
   
    Many thanks, Eibe. Highly appreciat your extensive help.
   
    Good luck to you.
   
    Kind regards,
    Andria
    On 6 Jun 2017 6:48 a.m., "Eibe Frank" <[hidden email]> wrote:
   
    >
    > > On 5/06/2017, at 10:27 PM, Andria Lan <[hidden email]> wrote:
    > >
    > > Thanks a lot for the extensive information. Finally, just wanted to
    > confirm the following issue:
    > >
    > > > GridSearch only works for exactly two parameters, and
    > CVParameterSelection cannot optimise nested parameters.
    > >
    > > Did you mean that MultiSearch works similar to GridSearch for the goal
    > of parameters optimisation. Right?
    >
    > Yes. GridSearch has the option to extend the grid if necessary, which
    > MultiSearch doesn?t have, but it can only deal with optimisation problems
    > involving exactly two parameters.
    >
    > Cheers,
    > Eibe
    >
    >
    > _______________________________________________
    > Wekalist mailing list
    > Send posts to: [hidden email]
    > List info and subscription status:
    https://list.waikato.ac.nz/ <https://list.waikato.ac.nz/>Mailing Lists at The University of Waikato <https://list.waikato.ac.nz/>
    list.waikato.ac.nz
    University Mailing Lists. ITS provides mailing list services for all Waikato University departments that want to use them. Lists can be maintained manually, or ...
   
   
   
    > mailman/listinfo/wekalist
    > List etiquette:
    http://www.cs.waikato.ac.nz/~ <http://www.cs.waikato.ac.nz/~>
    > ml/weka/mailinglist_etiquette.html
    >
    -------------- next part --------------
    An HTML attachment was scrubbed...
    URL: <http://list.waikato.ac.nz/pipermail/wekalist/attachments/20170607/0a9417a3/attachment-0001.html>
   
    ------------------------------
   
    Message: 2
    Date: Wed, 7 Jun 2017 16:47:54 +1200
    From: Eibe Frank <[hidden email]>
    To: "Weka machine learning workbench list."
            <[hidden email]>
    Subject: Re: [Wekalist] Weka-Spark
    Message-ID: <[hidden email]>
    Content-Type: text/plain; charset=utf-8
   
   
    > On 7/06/2017, at 8:34 AM, ENGMohammed kamal <[hidden email]> wrote:
    >
    > when i put sampling techniques only in weka spark job and remove it from weka spark evaluation job (no change in values ) .. why ?
   
    Can you be more specific about what you did?
   
    > i have more 3 questions
    > 1- if i have 10 classifiers and iwant to draw ROC curve for them for comparison ( with best high resolution )  , you told me befor this is not available in knowledge flow , how can i do this using explorer or experminter ?  note : i try normal weka o/p  it's
     not clear and most of line overlapped . with sympols like + & *
   
    The Explorer uses the same code as the KnowledgeFlow. Yes, you can use the KnowledgeFlow to plot ROC curves. You can also save the data for a plot into a file, using the Save? button in the plotting window. Then you can use an external tool to make a high-resolution
     plot.
   
    > 2-  i worked with vote classifiers vote ensemble in weka spark  , the output model for JRip ( contains 4 model) as i sliced my dataset to 4 partitions ( please see attached file) , but how can i know the best winning model between the 4 models by voting.
   
    Voting will use all the models. It?s does not use a single model for prediction.
   
    > 3-  if i reference my work in research paper for experiments which are done by weka -spark, it is enough these two references or i need more , specially for weka-spark ??
   
    > M.Hall, Mark, Eibe Frank, Geoffrey Holmes, Bernhard Pfahringer, Peter Reutemann, and Ian H. Witten. "The WEKA data mining software: an update." ACM SIGKDD explorations newsletter 11, no. 1 (2009): 10-18.
    >
    > [69] Eibe Frank, Mark A. Hall, and Ian H. Witten (2016). The WEKA Workbench. Online Appendix for "Data Mining: Practical Machine Learning Tools and Techniques", Morgan Kaufmann, Fourth Edition, 2016.
   
    That sounds fine.
   
    Cheers,
    Eibe
   
   
   
    ------------------------------
   
    Message: 3
    Date: Wed, 7 Jun 2017 20:22:59 +1200
    From: Eibe Frank <[hidden email]>
    To: "Weka machine learning workbench list."
            <[hidden email]>
    Subject: Re: [Wekalist] AutoWEKA on Windows
    Message-ID: <[hidden email]>
    Content-Type: text/plain; charset=utf-8
   
    I use WEKA on Windows 10 all the time, and Windows is the most popular platform for running WEKA. There must be something specific to your particular set-up that's causing problems.
   
    Cheers,
    Eibe
   
    > On 6 Jun 2017, at 21:32, Alexander Osherenko <[hidden email]> wrote:
    >
    > The problem in Weka installation on Win10 is RunWeka.bat. If I comment out the first line in it I can see that last line cant't be executed if called on the console from outside the Weka directory (works good if called from the Weka directory):
    >
    > javaw -classpath . RunWeka -i .\RunWeka.ini -w .\weka.jar -c default ""
    >
    > Consequently, although the Weka shortcut specifies the Run in directory, the target ..\Weka-3-8\RunWeka.bat default doesn't work. Specifying instead the target ..\Weka-3-8\debug_runWeka.bat (my own) that contains one line
    >
    > java -jar ...\Weka-3-8\weka.jar
    >
    > shows up Weka started from Win10 menu.
    >
    > Best, Alexander?
    >
    > 2017-06-05 12:43 GMT+01:00 Alexander Osherenko <[hidden email]>:
    > There are problems with paths on Win10.
    >
    > 1. It begins with installing Weka that can't start from menu although the shortcut seems to be OK: Target and Start in folder are correct, but RunWeka.bat doesn't work. Windows console flashes but doesn't start Weka -- I assume, Windows doesn't find RunWeka.bat
     or RunWeka.bat crashes.
    >
    > ?Whatsoever, ??I developed a batch ??workaround ?file that does exactly the same as the shortcut should:
    > set _JAVA_OPTIONS=-Djava.class.path=./weka.jar;
    > cd D:\toDeleteJDK\Weka-3-8
    > d:
    > D:\toDeleteJDK\Weka-3-8\RunWeka.bat default
    >
    > 2. This starts Weka and EVEN the Weka Explorer )). But there are other problems: The AutoWeka package is installed on the hard drive but not in Weka -- AutoWeka classes are not found what I can see on the Weka log, for example, instantiation of AutoWEKA classes
     fails. I added autoweka.jar manually to the first line of my batch file.
    >
    > _JAVA_OPTIONS=-Djava.class.path=./weka.jar;C:\Users\osher\wekafiles\packages\Auto-WEKA\autoweka.jar;
    >
    > and now Weka finds necessary AutoWeka classes but the path problem is not solved: I see no instantiation errors on the log anymore, but other error messages (AutoWeka complains about failures), for example,
    > Experiment - Failed to parse trajectory
    >
    > There are no errors on XP at all (neither with the Weka shortcut, nor with the missing autoweka.jar). I assume there are problems with incorrect settings in Win10, but the Weka and the AutoWeka reinstalls to update paths don't help.
    >
    > ?Best, Alexander
    > ?
    > 2017-06-05 5:30 GMT+01:00 Eibe Frank <[hidden email]>:
    > Yes, on my Mac, the following works:
    >
    > export _JAVA_OPTIONS="-Dsun.java.command=dummy -Djava.home=/Library/Java/JavaVirtualMachines/jdk1.8.0_102.jdk/Contents/Home/jre"
    >
    > Cheers,
    > Eibe
    >
    > PS: Interestingly, I had to set -Djava.home to .../jre to make WEKA work, but the /jre part can be omitted when using the JAVA_HOME environment variable instead.
    >
    >
    > > On 5 Jun 2017, at 07:08, Peter Reutemann <[hidden email]> wrote:
    > >
    > > On June 5, 2017 2:57:55 AM GMT+12:00, Alexander Osherenko <[hidden email]> wrote:
    > > Is it possible to set several variables in JAVA_OPTIONS, for example,
    > > to set _JAVA_OPTIONS to sun.java.command=var1 and java.home=var2?
    > >
    > > If I specify string -Dsun.java.command=var1;java.home=var2, it is not parsed correctly in Systeminfo.
    > >
    > > Cheers, Alexander
    > >
    > > 2017-06-03 8:59 GMT+01:00 Alexander Osherenko <[hidden email]>:
    > > Got it, thanx. It is also possible to set the sun.java.command in Systeminfo by setting
    > > _JAVA_OPTIONS=-Dsun.java.command="dummy".
    > >
    > > Best, Alexander
    > >
    > > 2017-06-03 5:44 GM??T+01:00 Eibe Frank <[hidden email]>:
    > >
    > > Yes, modifying RunWeka.bat should work too.
    > >
    > > The advantage of using an environment variable is that you won't have to fix this problem again when you install a different version of WEKA. Also, it works even if you don't have admin rights. A correct setting for JAVA_HOME may also be useful for other
     applications, so from that point of view it may also make sense to set this globally in an environment variable.
    > >
    > > The advantage of using _JAVA_OPTIONS in particular is that any settings specified in this variable take precedence over other settings provided for the java command, see
    > >
    > >
    https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts> <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts>
    Difference between _JAVA_OPTIONS JAVA ...
     - Stack Overflow <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts>
    stackoverflow.com
    I thought it would be great to have a comparison between _JAVA_OPTIONS and JAVA_TOOL_OPTIONS. I have been searching a bit for one, but I cannot find anything, so I ...
   
   
   
    > >
    > > Cheers,
    > > Eibe
    > >
    > > > On 3 Jun 2017, at 16:08, Alexander Osherenko <[hidden email]> wrote:
    > > >
    > > > I wonder why you are talking about this variable. Is it the only way to set java.home variable in Weka or I can set it in RunWeka.bat?
    > > >
    > > > ?Cheers, Alexander
    > > > ?
    > > > 2017-06-02 23:10 GMT+01:00 Eibe Frank <[hidden email]>:
    > > > Try setting a new variable
    > > >
    > > >   _JAVA_OPTIONS
    > > >
    > > > as a Windows environment variable and give it the value
    > > >
    > > >   -Djava.home=...
    > > >
    > > > Cheers,
    > > > Eibe
    > > >
    > > >
    > > >
    > > > Sent from my iPad
    > > > On 3/06/2017, at 7:15 AM, Alexander Osherenko <[hidden email]> wrote:
    > > >
    > > >> I tried to find the problem and didn't. Maybe you remember I have Weka 3-8-1 and AutoWEKA installed and they are running without any problems on XP. That's why I have to seek further.
    > > >>
    > > >> Let us please compare the settings in Weka Systeminfo. I found some differences, for example, in java.home and sun.java.command. Can I set these variables in RunWeka.bat?
    > > >>
    > > >> 2017-06-02 9:24 GMT+01:00 Michael Hall <[hidden email]>:
    > > >> > On Jun 1, 2017, at 8:48 PM, Eibe Frank <[hidden email]> wrote:
    > > >> >
    > > >> > With Oracle JVMs, you can use the _JAVA_OPTIONS environment variable to pass such command-line parameters to the JVM. This works on all platforms.
    > > >>
    > > >>
    > > >> I wasn?t aware of that one. It should work for this then, I would think. Unless Java itself sets java.home and overrides the one you specify.
    > > >>
    > > >> I didn?t entirely follow this thread and I am not sure why it is necessary. A jvm to run the application is obviously being found and if the same one is used to set the java.home property you would think AutoWeka would work with that? Possibly some localization
     issue as the original poster indicated. Then setting the property could work.
    > > >>
    > > >> The other possibility would be that it is the same situation as on OS X. For that the ?java? executable command is not included in the Weka application embedded jre. So AutoWeka can?t find it there and fails to Runtime exec java. I still believe this
     is the behavior of the appbundler version that is being used to create the application bundle. Probably just to save some space the executables are omitted. I also still think a fix there would be for whatever is used for the application build process to create
     a ?bin' directory in the embedded jre and copy in the ?java? executable command. This did in fact work for me, doing it myself, to get AutoWeka going on OS X.
    > > >>
    > > >> For what this has to do with Windows, if you try to override java.home I would make sure you are pointing it at a jre/jdk that includes bin/java or bin/java.exe.
    > > >>
    > > >> Michael Hall
    > > >>
    > > >>
    > > >> _______________________________________________
    > > >> Wekalist mailing list
    > > >> Send posts to: [hidden email]
    > > >> List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > > >> List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    > > >>
    > > >> _______________________________________________
    > > >> Wekalist mailing list
    > > >> Send posts to: [hidden email]
    > > >> List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > > >> List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    > > >
    > > > _______________________________________________
    > > > Wekalist mailing list
    > > > Send posts to: [hidden email]
    > > > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > > > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    > > >
    > > >
    > > > _______________________________________________
    > > > Wekalist mailing list
    > > > Send posts to: [hidden email]
    > > > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > > > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    > >
    > > _______________________________________________
    > > Wekalist mailing list
    > > Send posts to: [hidden email]
    > > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    > >
    > >
    > >
    > > Separate the options with blanks instead of semicolons. This variable acts like a partial command-line string, which gets added to the actual command-line to start the jvm.
    > >
    > > Cheers, Peter
    > > --
    > > Peter Reutemann
    > > Dept. of Computer Science
    > > University of Waikato, NZ
    > > +64 (7) 858-5174
    > >
    http://www.cms.waikato.ac.nz/~fracpete/ <http://www.cms.waikato.ac.nz/~fracpete/>home | fracpete <http://www.cms.waikato.ac.nz/~fracpete/>
    www.cms.waikato.ac.nz
    Open-source advocate and programmer by heart, specialized in workflow engines, machine learning and data mining.
   
   
   
    > >
    http://www.data-mining.co.nz <http://www.data-mining.co.nz>Data Mining : University of Waikato <http://www.data-mining.co.nz/>
    www.data-mining.co.nz
    The University of Waikato is the home of several open-source software packages for data mining: WEKA - batch learning (preprocessing, classification, regression ...
   
   
   
    > > _______________________________________________
    > > Wekalist mailing list
    > > Send posts to: [hidden email]
    > > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    >
    > _______________________________________________
    > Wekalist mailing list
    > Send posts to: [hidden email]
    > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    >
    >
    > _______________________________________________
    > Wekalist mailing list
    > Send posts to: [hidden email]
    > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
   
   
   
    ------------------------------
   
    _______________________________________________
    Wekalist mailing list
    [hidden email]
    https://list.waikato.ac.nz/mailman/listinfo/wekalistWekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
   
   
    End of Wekalist Digest, Vol 172, Issue 24
    *****************************************
   
   
   
   
   
    _______________________________________________
    Wekalist mailing list
    Send posts to: [hidden email]
    List info and subscription status: 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]
List info and subscription status: 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: Weka-Spark

MOHAMMED KAMAL
In reply to this post by MOHAMMED KAMAL

Hi Mark,

yes for final model evaluation 
Mark you comment only on first question
do you have answers for the next two questions
cheers

M.kamal







From: [hidden email] <[hidden email]> on behalf of [hidden email] <[hidden email]>
Sent: Monday, June 19, 2017 1:32 AM
To: [hidden email]
Subject: Wekalist Digest, Vol 172, Issue 64
 
Send Wekalist mailing list submissions to
        [hidden email]

To subscribe or unsubscribe via the World Wide Web, visit
        https://list.waikato.ac.nz/mailman/listinfo/wekalist


or, via email, send a message with subject or body 'help' to
        [hidden email]

You can reach the person managing the list at
        [hidden email]

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Wekalist digest..."


Today's Topics:

   1. Re: Regression with M5p (Eibe Frank)
   2. Re: Weka-Spark (Mark Hall)


----------------------------------------------------------------------

Message: 1
Date: Mon, 19 Jun 2017 10:23:19 +1200
From: Eibe Frank <[hidden email]>
To: "Weka machine learning workbench list."
        <[hidden email]>
Subject: Re: [Wekalist] Regression with M5p
Message-ID: <[hidden email]>
Content-Type: text/plain; charset=us-ascii

The classifyInstance() method is the correct method to use. Why is a small error problematic?

Cheers,
Eibe

> On 18/06/2017, at 8:49 PM, Gaetano <[hidden email]> wrote:
>
> Hi Eibe
> I would like to ask you to make numerical prediction for each instance, the
> only available statement is "classifyInstance" istruction (that i write
> before)?
> Because afterwards, when I calculate the rmse, I get very low values
> (example 2.325554E-5) and I can not understand the reason.
> Thanks
>
>
>
> --
> View this message in context: http://weka.8497.n7.nabble.com/Regression-with-M5p-tp40847p40980.html


> Sent from the WEKA mailing list archive at Nabble.com.
> _______________________________________________
> Wekalist mailing list
> Send posts to: [hidden email]
> List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist


> List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html





------------------------------

Message: 2
Date: Mon, 19 Jun 2017 10:32:47 +1200
From: Mark Hall <[hidden email]>
To: "Weka machine learning workbench list."
        <[hidden email]>
Subject: Re: [Wekalist] Weka-Spark
Message-ID: <[hidden email]>
Content-Type: text/plain;       charset="UTF-8"

I don't think there is a problem using SMOTE in the evaluation job, as only the partial training folds will be affected by the SMOTE process. What is the purpose of running the classifier job before the evaluation job? Is this just to create a final model?

Cheers,
Mark.


On 19/06/17, 9:18 AM, "ENGMohammed kamal" <[hidden email] on behalf of [hidden email]> wrote:

    Hi Eibe,
    >
    > when i put
     sampling techniques only in weka spark job and remove
     it from weka spark evaluation job (no change in values
     ) .. why ?
   
    Q1. Can you be more specific about what you did?
    __________________________________________
   
   
    i put  the following sequence :
    arff header spark job (4 data slices)  ---> randomly shuffle data spark job  ----->
     i add  sampling technique (SMOTE for example) + weka classifier spark job --------->   weka classifier evaluation without sampling (as you recommensd me before)
   
    nothing change in accuracy and f-measure ??
    if i put sampling also in weka classification evaluation spark job i get higher accuracy and f-measure
    note: you recommend me to do not that
    ________________________________________
    Q2. if i compare 10 classifiers without sampling and with sampling techniques (class balancer-ROS-RUS-Smote)  on imbalanced dataset what measure should i put in the table for comparison ?
    is it enough to put accuracy , average f-measure and f measure for minority class ? what kind of graphics will be demonstrative for academic papers ?
    ------------------------------------------------------------------------------
    Q3, i asked about  ROC curve to compare these classifiers
    your reply : The Explorer uses the same code as the KnowledgeFlow.
     Yes, you can use the KnowledgeFlow to plot ROC curves. You can also save the data for a plot into a file, using the Save? button in the plotting window. Then you can use an external tool to make a high-resolution plot.
    please can you give me further information, i didn't understand this point
    do you mean to make a plot for every classifier and save it separately and then aggregate all with another tool?   which tool do you recommend ?   how can i do this? 
    i have 10 classifiers ( without - with CB-RUS-ROS-SMOTE)
    50 readings , so will be huge to compare figures
    i think i can give a figure compare each classifier with sampling with / without sampling techniques using ROC curve?
    or do you recommend to make bar chart for F-Measure in minority class only
    i want your recommendation
   
   
    cheers
    M.kamal
   
   
   
   
   
   
   
   
   
   
    ________________________________________
    From: [hidden email] <[hidden email]> on behalf of [hidden email] <[hidden email]>
    Sent: Wednesday, June 7, 2017 11:23 AM
    To: [hidden email]
    Subject: Wekalist Digest, Vol 172, Issue 24
   
   
    Send Wekalist mailing list submissions to
            [hidden email]
   
    To subscribe or unsubscribe via the World Wide Web, visit
           
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>




    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    or, via email, send a message with subject or body 'help' to
            [hidden email]
   
    You can reach the person managing the list at
            [hidden email]
   
    When replying, please edit your Subject line so it is more specific
    than "Re: Contents of Wekalist digest..."
   
   
    Today's Topics:
   
       1. Re: LSA implementation (Andria Lan)
       2. Re: Weka-Spark (Eibe Frank)
       3. Re: AutoWEKA on Windows (Eibe Frank)
   
   
    ----------------------------------------------------------------------
   
    Message: 1
    Date: Wed, 7 Jun 2017 09:36:59 +0800
    From: Andria Lan <[hidden email]>
    To: "Weka machine learning workbench list."
            <[hidden email]>
    Subject: Re: [Wekalist] LSA implementation
    Message-ID:
            <CAHM6HB0gqV2YGySomFSd0ToMObVm0Vg7KXDWEaM-W7vhekS=[hidden email]>
    Content-Type: text/plain; charset="utf-8"
   
    Many thanks, Eibe. Highly appreciat your extensive help.
   
    Good luck to you.
   
    Kind regards,
    Andria
    On 6 Jun 2017 6:48 a.m., "Eibe Frank" <[hidden email]> wrote:
   
    >
    > > On 5/06/2017, at 10:27 PM, Andria Lan <[hidden email]> wrote:
    > >
    > > Thanks a lot for the extensive information. Finally, just wanted to
    > confirm the following issue:
    > >
    > > > GridSearch only works for exactly two parameters, and
    > CVParameterSelection cannot optimise nested parameters.
    > >
    > > Did you mean that MultiSearch works similar to GridSearch for the goal
    > of parameters optimisation. Right?
    >
    > Yes. GridSearch has the option to extend the grid if necessary, which
    > MultiSearch doesn?t have, but it can only deal with optimisation problems
    > involving exactly two parameters.
    >
    > Cheers,
    > Eibe
    >
    >
    > _______________________________________________
    > Wekalist mailing list
    > Send posts to: [hidden email]
    > List info and subscription status:
    https://list.waikato.ac.nz/ <https://list.waikato.ac.nz/>Mailing Lists at The University of Waikato <https://list.waikato.ac.nz/>


    list.waikato.ac.nz
    University Mailing Lists. ITS provides mailing list services for all Waikato University departments that want to use them. Lists can be maintained manually, or ...
   
   
   
    > mailman/listinfo/wekalist
    > List etiquette:
    http://www.cs.waikato.ac.nz/~ <http://www.cs.waikato.ac.nz/~>
    > ml/weka/mailinglist_etiquette.html
    >
    -------------- next part --------------
    An HTML attachment was scrubbed...
    URL: <http://list.waikato.ac.nz/pipermail/wekalist/attachments/20170607/0a9417a3/attachment-0001.html>
   
    ------------------------------
   
    Message: 2
    Date: Wed, 7 Jun 2017 16:47:54 +1200
    From: Eibe Frank <[hidden email]>
    To: "Weka machine learning workbench list."
            <[hidden email]>
    Subject: Re: [Wekalist] Weka-Spark
    Message-ID: <[hidden email]>
    Content-Type: text/plain; charset=utf-8
   
   
    > On 7/06/2017, at 8:34 AM, ENGMohammed kamal <[hidden email]> wrote:
    >
    > when i put sampling techniques only in weka spark job and remove it from weka spark evaluation job (no change in values ) .. why ?
   
    Can you be more specific about what you did?
   
    > i have more 3 questions
    > 1- if i have 10 classifiers and iwant to draw ROC curve for them for comparison ( with best high resolution )  , you told me befor this is not available in knowledge flow , how can i do this using explorer or experminter ?  note : i try normal weka o/p  it's
     not clear and most of line overlapped . with sympols like + & *
   
    The Explorer uses the same code as the KnowledgeFlow. Yes, you can use the KnowledgeFlow to plot ROC curves. You can also save the data for a plot into a file, using the Save? button in the plotting window. Then you can use an external tool to make a high-resolution
     plot.
   
    > 2-  i worked with vote classifiers vote ensemble in weka spark  , the output model for JRip ( contains 4 model) as i sliced my dataset to 4 partitions ( please see attached file) , but how can i know the best winning model between the 4 models by voting.
   
    Voting will use all the models. It?s does not use a single model for prediction.
   
    > 3-  if i reference my work in research paper for experiments which are done by weka -spark, it is enough these two references or i need more , specially for weka-spark ??
   
    > M.Hall, Mark, Eibe Frank, Geoffrey Holmes, Bernhard Pfahringer, Peter Reutemann, and Ian H. Witten. "The WEKA data mining software: an update." ACM SIGKDD explorations newsletter 11, no. 1 (2009): 10-18.
    >
    > [69] Eibe Frank, Mark A. Hall, and Ian H. Witten (2016). The WEKA Workbench. Online Appendix for "Data Mining: Practical Machine Learning Tools and Techniques", Morgan Kaufmann, Fourth Edition, 2016.
   
    That sounds fine.
   
    Cheers,
    Eibe
   
   
   
    ------------------------------
   
    Message: 3
    Date: Wed, 7 Jun 2017 20:22:59 +1200
    From: Eibe Frank <[hidden email]>
    To: "Weka machine learning workbench list."
            <[hidden email]>
    Subject: Re: [Wekalist] AutoWEKA on Windows
    Message-ID: <[hidden email]>
    Content-Type: text/plain; charset=utf-8
   
    I use WEKA on Windows 10 all the time, and Windows is the most popular platform for running WEKA. There must be something specific to your particular set-up that's causing problems.
   
    Cheers,
    Eibe
   
    > On 6 Jun 2017, at 21:32, Alexander Osherenko <[hidden email]> wrote:
    >
    > The problem in Weka installation on Win10 is RunWeka.bat. If I comment out the first line in it I can see that last line cant't be executed if called on the console from outside the Weka directory (works good if called from the Weka directory):
    >
    > javaw -classpath . RunWeka -i .\RunWeka.ini -w .\weka.jar -c default ""
    >
    > Consequently, although the Weka shortcut specifies the Run in directory, the target ..\Weka-3-8\RunWeka.bat default doesn't work. Specifying instead the target ..\Weka-3-8\debug_runWeka.bat (my own) that contains one line
    >
    > java -jar ...\Weka-3-8\weka.jar
    >
    > shows up Weka started from Win10 menu.
    >
    > Best, Alexander?
    >
    > 2017-06-05 12:43 GMT+01:00 Alexander Osherenko <[hidden email]>:
    > There are problems with paths on Win10.
    >
    > 1. It begins with installing Weka that can't start from menu although the shortcut seems to be OK: Target and Start in folder are correct, but RunWeka.bat doesn't work. Windows console flashes but doesn't start Weka -- I assume, Windows doesn't find RunWeka.bat
     or RunWeka.bat crashes.
    >
    > ?Whatsoever, ??I developed a batch ??workaround ?file that does exactly the same as the shortcut should:
    > set _JAVA_OPTIONS=-Djava.class.path=./weka.jar;
    > cd D:\toDeleteJDK\Weka-3-8
    > d:
    > D:\toDeleteJDK\Weka-3-8\RunWeka.bat default
    >
    > 2. This starts Weka and EVEN the Weka Explorer )). But there are other problems: The AutoWeka package is installed on the hard drive but not in Weka -- AutoWeka classes are not found what I can see on the Weka log, for example, instantiation of AutoWEKA classes
     fails. I added autoweka.jar manually to the first line of my batch file.
    >
    > _JAVA_OPTIONS=-Djava.class.path=./weka.jar;C:\Users\osher\wekafiles\packages\Auto-WEKA\autoweka.jar;
    >
    > and now Weka finds necessary AutoWeka classes but the path problem is not solved: I see no instantiation errors on the log anymore, but other error messages (AutoWeka complains about failures), for example,
    > Experiment - Failed to parse trajectory
    >
    > There are no errors on XP at all (neither with the Weka shortcut, nor with the missing autoweka.jar). I assume there are problems with incorrect settings in Win10, but the Weka and the AutoWeka reinstalls to update paths don't help.
    >
    > ?Best, Alexander
    > ?
    > 2017-06-05 5:30 GMT+01:00 Eibe Frank <[hidden email]>:
    > Yes, on my Mac, the following works:
    >
    > export _JAVA_OPTIONS="-Dsun.java.command=dummy -Djava.home=/Library/Java/JavaVirtualMachines/jdk1.8.0_102.jdk/Contents/Home/jre"
    >
    > Cheers,
    > Eibe
    >
    > PS: Interestingly, I had to set -Djava.home to .../jre to make WEKA work, but the /jre part can be omitted when using the JAVA_HOME environment variable instead.
    >
    >
    > > On 5 Jun 2017, at 07:08, Peter Reutemann <[hidden email]> wrote:
    > >
    > > On June 5, 2017 2:57:55 AM GMT+12:00, Alexander Osherenko <[hidden email]> wrote:
    > > Is it possible to set several variables in JAVA_OPTIONS, for example,
    > > to set _JAVA_OPTIONS to sun.java.command=var1 and java.home=var2?
    > >
    > > If I specify string -Dsun.java.command=var1;java.home=var2, it is not parsed correctly in Systeminfo.
    > >
    > > Cheers, Alexander
    > >
    > > 2017-06-03 8:59 GMT+01:00 Alexander Osherenko <[hidden email]>:
    > > Got it, thanx. It is also possible to set the sun.java.command in Systeminfo by setting
    > > _JAVA_OPTIONS=-Dsun.java.command="dummy".
    > >
    > > Best, Alexander
    > >
    > > 2017-06-03 5:44 GM??T+01:00 Eibe Frank <[hidden email]>:
    > >
    > > Yes, modifying RunWeka.bat should work too.
    > >
    > > The advantage of using an environment variable is that you won't have to fix this problem again when you install a different version of WEKA. Also, it works even if you don't have admin rights. A correct setting for JAVA_HOME may also be useful for other
     applications, so from that point of view it may also make sense to set this globally in an environment variable.
    > >
    > > The advantage of using _JAVA_OPTIONS in particular is that any settings specified in this variable take precedence over other settings provided for the java command, see
    > >
    > >
    https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts> <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts>
    Difference between _JAVA_OPTIONS JAVA ...
     - Stack Overflow <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts>
    stackoverflow.com
    I thought it would be great to have a comparison between _JAVA_OPTIONS and JAVA_TOOL_OPTIONS. I have been searching a bit for one, but I cannot find anything, so I ...
   
   
   
    > >
    > > Cheers,
    > > Eibe
    > >
    > > > On 3 Jun 2017, at 16:08, Alexander Osherenko <[hidden email]> wrote:
    > > >
    > > > I wonder why you are talking about this variable. Is it the only way to set java.home variable in Weka or I can set it in RunWeka.bat?
    > > >
    > > > ?Cheers, Alexander
    > > > ?
    > > > 2017-06-02 23:10 GMT+01:00 Eibe Frank <[hidden email]>:
    > > > Try setting a new variable
    > > >
    > > >   _JAVA_OPTIONS
    > > >
    > > > as a Windows environment variable and give it the value
    > > >
    > > >   -Djava.home=...
    > > >
    > > > Cheers,
    > > > Eibe
    > > >
    > > >
    > > >
    > > > Sent from my iPad
    > > > On 3/06/2017, at 7:15 AM, Alexander Osherenko <[hidden email]> wrote:
    > > >
    > > >> I tried to find the problem and didn't. Maybe you remember I have Weka 3-8-1 and AutoWEKA installed and they are running without any problems on XP. That's why I have to seek further.
    > > >>
    > > >> Let us please compare the settings in Weka Systeminfo. I found some differences, for example, in java.home and sun.java.command. Can I set these variables in RunWeka.bat?
    > > >>
    > > >> 2017-06-02 9:24 GMT+01:00 Michael Hall <[hidden email]>:
    > > >> > On Jun 1, 2017, at 8:48 PM, Eibe Frank <[hidden email]> wrote:
    > > >> >
    > > >> > With Oracle JVMs, you can use the _JAVA_OPTIONS environment variable to pass such command-line parameters to the JVM. This works on all platforms.
    > > >>
    > > >>
    > > >> I wasn?t aware of that one. It should work for this then, I would think. Unless Java itself sets java.home and overrides the one you specify.
    > > >>
    > > >> I didn?t entirely follow this thread and I am not sure why it is necessary. A jvm to run the application is obviously being found and if the same one is used to set the java.home property you would think AutoWeka would work with that? Possibly some localization
     issue as the original poster indicated. Then setting the property could work.
    > > >>
    > > >> The other possibility would be that it is the same situation as on OS X. For that the ?java? executable command is not included in the Weka application embedded jre. So AutoWeka can?t find it there and fails to Runtime exec java. I still believe this
     is the behavior of the appbundler version that is being used to create the application bundle. Probably just to save some space the executables are omitted. I also still think a fix there would be for whatever is used for the application build process to create
     a ?bin' directory in the embedded jre and copy in the ?java? executable command. This did in fact work for me, doing it myself, to get AutoWeka going on OS X.
    > > >>
    > > >> For what this has to do with Windows, if you try to override java.home I would make sure you are pointing it at a jre/jdk that includes bin/java or bin/java.exe.
    > > >>
    > > >> Michael Hall
    > > >>
    > > >>
    > > >> _______________________________________________
    > > >> Wekalist mailing list
    > > >> Send posts to: [hidden email]
    > > >> List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > > >> List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    > > >>
    > > >> _______________________________________________
    > > >> Wekalist mailing list
    > > >> Send posts to: [hidden email]
    > > >> List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > > >> List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    > > >
    > > > _______________________________________________
    > > > Wekalist mailing list
    > > > Send posts to: [hidden email]
    > > > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > > > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    > > >
    > > >
    > > > _______________________________________________
    > > > Wekalist mailing list
    > > > Send posts to: [hidden email]
    > > > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > > > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    > >
    > > _______________________________________________
    > > Wekalist mailing list
    > > Send posts to: [hidden email]
    > > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    > >
    > >
    > >
    > > Separate the options with blanks instead of semicolons. This variable acts like a partial command-line string, which gets added to the actual command-line to start the jvm.
    > >
    > > Cheers, Peter
    > > --
    > > Peter Reutemann
    > > Dept. of Computer Science
    > > University of Waikato, NZ
    > > +64 (7) 858-5174
    > >
    http://www.cms.waikato.ac.nz/~fracpete/ <http://www.cms.waikato.ac.nz/~fracpete/>home | fracpete <http://www.cms.waikato.ac.nz/~fracpete/>
    www.cms.waikato.ac.nz
    Open-source advocate and programmer by heart, specialized in workflow engines, machine learning and data mining.
   
   
   
    > >
    http://www.data-mining.co.nz <http://www.data-mining.co.nz>Data Mining : University of Waikato <http://www.data-mining.co.nz/>
    www.data-mining.co.nz
    The University of Waikato is the home of several open-source software packages for data mining: WEKA - batch learning (preprocessing, classification, regression ...
   
   
   
    > > _______________________________________________
    > > Wekalist mailing list
    > > Send posts to: [hidden email]
    > > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    >
    > _______________________________________________
    > Wekalist mailing list
    > Send posts to: [hidden email]
    > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
    >
    >
    > _______________________________________________
    > Wekalist mailing list
    > Send posts to: [hidden email]
    > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
   
   
   
    ------------------------------
   
    _______________________________________________
    Wekalist mailing list
    [hidden email]
    https://list.waikato.ac.nz/mailman/listinfo/wekalistWekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
   
   
    End of Wekalist Digest, Vol 172, Issue 24
    *****************************************
   
   
   
   
   
    _______________________________________________
    Wekalist mailing list
    Send posts to: [hidden email]
    List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
    List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
   




------------------------------

_______________________________________________
Wekalist mailing list
[hidden email]
https://list.waikato.ac.nz/mailman/listinfo/wekalist


End of Wekalist Digest, Vol 172, Issue 64
*****************************************

_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: 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: Weka-Spark

Mark Hall
In reply to this post by MOHAMMED KAMAL
If you are dealing with imbalanced class distributions, then measures such as AUC and AUPRC are more appropriate than accuracy.

You can save the data for each ROC curve in the Explorer from the scatter plot that pops up when you visualize a threshold curve. There are various tools that can handle plotting more than one series of data – Excel is one; gnuplot is another. It is beyond the scope of this forum to explain how to use such tools.

Cheers,
Mark.

On 19/06/17, 11:20 AM, "ENGMohammed kamal" <[hidden email] on behalf of [hidden email]> wrote:

    Hi Mark,
    yes for final model evaluation
    Mark you comment only on first question
    do you have answers for the next two questions
    cheers
    M.kamal
   
   
    ________________________________________
   
   
   
   
    ________________________________________
    From: [hidden email] <[hidden email]> on behalf of [hidden email] <[hidden email]>
    Sent: Monday, June 19, 2017 1:32 AM
    To: [hidden email]
    Subject: Wekalist Digest, Vol 172, Issue 64
   
   
    Send Wekalist mailing list submissions to
            [hidden email]
   
    To subscribe or unsubscribe via the World Wide Web, visit
           
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    or, via email, send a message with subject or body 'help' to
            [hidden email]
   
    You can reach the person managing the list at
            [hidden email]
   
    When replying, please edit your Subject line so it is more specific
    than "Re: Contents of Wekalist digest..."
   
   
    Today's Topics:
   
       1. Re: Regression with M5p (Eibe Frank)
       2. Re: Weka-Spark (Mark Hall)
   
   
    ----------------------------------------------------------------------
   
    Message: 1
    Date: Mon, 19 Jun 2017 10:23:19 +1200
    From: Eibe Frank <[hidden email]>
    To: "Weka machine learning workbench list."
            <[hidden email]>
    Subject: Re: [Wekalist] Regression with M5p
    Message-ID: <[hidden email]>
    Content-Type: text/plain; charset=us-ascii
   
    The classifyInstance() method is the correct method to use. Why is a small error problematic?
   
    Cheers,
    Eibe
   
    > On 18/06/2017, at 8:49 PM, Gaetano <[hidden email]> wrote:
    >
    > Hi Eibe
    > I would like to ask you to make numerical prediction for each instance, the
    > only available statement is "classifyInstance" istruction (that i write
    > before)?
    > Because afterwards, when I calculate the rmse, I get very low values
    > (example 2.325554E-5) and I can not understand the reason.
    > Thanks
    >
    >
    >
    > --
    > View this message in context:
    http://weka.8497.n7.nabble.com/Regression-with-M5p-tp40847p40980.html <http://weka.8497.n7.nabble.com/Regression-with-M5p-tp40847p40980.html>WEKA - Regression with M5p <http://weka.8497.n7.nabble.com/Regression-with-M5p-tp40847p40980.html>
    weka.8497.n7.nabble.com
    Regression with M5p. Hi I have a test set and a training set and I should perform a numeric prediction of the class label using M5p. How it is possible to invoke m5p using java weka api? I tried this...
   
   
   
    > Sent from the WEKA mailing list archive at Nabble.com.
    > _______________________________________________
    > Wekalist mailing list
    > Send posts to: [hidden email]
    > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
    www.cs.waikato.ac.nz
    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
   
   
   
    ------------------------------
   
    Message: 2
    Date: Mon, 19 Jun 2017 10:32:47 +1200
    From: Mark Hall <[hidden email]>
    To: "Weka machine learning workbench list."
            <[hidden email]>
    Subject: Re: [Wekalist] Weka-Spark
    Message-ID: <[hidden email]>
    Content-Type: text/plain;       charset="UTF-8"
   
    I don't think there is a problem using SMOTE in the evaluation job, as only the partial training folds will be affected by the SMOTE process. What is the purpose of running the classifier job before the evaluation job? Is this just to create a final model?
   
    Cheers,
    Mark.
   
   
    On 19/06/17, 9:18 AM, "ENGMohammed kamal" <[hidden email] on behalf of [hidden email]> wrote:
   
        Hi Eibe,
        >
        > when i put
         sampling techniques only in weka spark job and remove
         it from weka spark evaluation job (no change in values
         ) .. why ?
       
        Q1. Can you be more specific about what you did?
        __________________________________________
       
       
        i put  the following sequence :
        arff header spark job (4 data slices)  ---> randomly shuffle data spark job  ----->
         i add  sampling technique (SMOTE for example) + weka classifier spark job --------->   weka classifier evaluation without sampling (as you recommensd me before)
       
        nothing change in accuracy and f-measure ??
        if i put sampling also in weka classification evaluation spark job i get higher accuracy and f-measure
   
        note: you recommend me to do not that
        ________________________________________
        Q2. if i compare 10 classifiers without sampling and with sampling techniques (class balancer-ROS-RUS-Smote)  on imbalanced dataset what measure should i put in the table for comparison ?
        is it enough to put accuracy , average f-measure and f measure for minority class ? what kind of graphics will be demonstrative for academic papers ?
   
        ------------------------------------------------------------------------------
        Q3, i asked about  ROC curve to compare these classifiers
        your reply : The Explorer uses the same code as the KnowledgeFlow.
         Yes, you can use the KnowledgeFlow to plot ROC curves. You can also save the data for a plot into a file, using the Save? button in the plotting window. Then you can use an external tool to make a high-resolution plot.
        please can you give me further information, i didn't understand this point
        do you mean to make a plot for every classifier and save it separately and then aggregate all with another tool?   which tool do you recommend ?   how can i do this?
   
        i have 10 classifiers ( without - with CB-RUS-ROS-SMOTE)
        50 readings , so will be huge to compare figures
        i think i can give a figure compare each classifier with sampling with / without sampling techniques using ROC curve?
   
        or do you recommend to make bar chart for F-Measure in minority class only
        i want your recommendation
       
       
        cheers
        M.kamal
       
       
       
       
       
       
       
       
       
       
        ________________________________________
        From: [hidden email] <[hidden email]> on behalf of [hidden email] <[hidden email]>
        Sent: Wednesday, June 7, 2017 11:23 AM
        To: [hidden email]
        Subject: Wekalist Digest, Vol 172, Issue 24
       
       
        Send Wekalist mailing list submissions to
                [hidden email]
       
        To subscribe or unsubscribe via the World Wide Web, visit
               
       
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist> <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato
     <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
    Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
    Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        or, via email, send a message with subject or body 'help' to
                [hidden email]
       
        You can reach the person managing the list at
                [hidden email]
       
        When replying, please edit your Subject line so it is more specific
        than "Re: Contents of Wekalist digest..."
       
       
        Today's Topics:
       
           1. Re: LSA implementation (Andria Lan)
           2. Re: Weka-Spark (Eibe Frank)
           3. Re: AutoWEKA on Windows (Eibe Frank)
       
       
        ----------------------------------------------------------------------
       
        Message: 1
        Date: Wed, 7 Jun 2017 09:36:59 +0800
        From: Andria Lan <[hidden email]>
        To: "Weka machine learning workbench list."
                <[hidden email]>
        Subject: Re: [Wekalist] LSA implementation
        Message-ID:
                <CAHM6HB0gqV2YGySomFSd0ToMObVm0Vg7KXDWEaM-W7vhekS=[hidden email]>
        Content-Type: text/plain; charset="utf-8"
       
        Many thanks, Eibe. Highly appreciat your extensive help.
       
        Good luck to you.
       
        Kind regards,
        Andria
        On 6 Jun 2017 6:48 a.m., "Eibe Frank" <[hidden email]> wrote:
       
        >
        > > On 5/06/2017, at 10:27 PM, Andria Lan <[hidden email]> wrote:
        > >
        > > Thanks a lot for the extensive information. Finally, just wanted to
        > confirm the following issue:
        > >
        > > > GridSearch only works for exactly two parameters, and
        > CVParameterSelection cannot optimise nested parameters.
        > >
        > > Did you mean that MultiSearch works similar to GridSearch for the goal
        > of parameters optimisation. Right?
        >
        > Yes. GridSearch has the option to extend the grid if necessary, which
        > MultiSearch doesn?t have, but it can only deal with optimisation problems
        > involving exactly two parameters.
        >
        > Cheers,
        > Eibe
        >
        >
        > _______________________________________________
        > Wekalist mailing list
        > Send posts to: [hidden email]
        > List info and subscription status:
       
    https://list.waikato.ac.nz/ <https://list.waikato.ac.nz/> <https://list.waikato.ac.nz/>Mailing Lists at The University of Waikato <https://list.waikato.ac.nz/>
    Mailing Lists at The University of Waikato <https://list.waikato.ac.nz/>
    list.waikato.ac.nz
    University Mailing Lists. ITS provides mailing list services for all Waikato University departments that want to use them. Lists can be maintained manually, or ...
   
   
   
        list.waikato.ac.nz
        University Mailing Lists. ITS provides mailing list services for all Waikato University departments that want to use them. Lists can be maintained manually, or ...
       
       
       
        > mailman/listinfo/wekalist
        > List etiquette:
        http://www.cs.waikato.ac.nz/~ <http://www.cs.waikato.ac.nz/~>
        > ml/weka/mailinglist_etiquette.html
        >
        -------------- next part --------------
        An HTML attachment was scrubbed...
        URL: <http://list.waikato.ac.nz/pipermail/wekalist/attachments/20170607/0a9417a3/attachment-0001.html>
       
        ------------------------------
       
        Message: 2
        Date: Wed, 7 Jun 2017 16:47:54 +1200
        From: Eibe Frank <[hidden email]>
        To: "Weka machine learning workbench list."
                <[hidden email]>
        Subject: Re: [Wekalist] Weka-Spark
        Message-ID: <[hidden email]>
        Content-Type: text/plain; charset=utf-8
       
       
        > On 7/06/2017, at 8:34 AM, ENGMohammed kamal <[hidden email]> wrote:
        >
        > when i put sampling techniques only in weka spark job and remove it from weka spark evaluation job (no change in values ) .. why ?
       
        Can you be more specific about what you did?
       
        > i have more 3 questions
        > 1- if i have 10 classifiers and iwant to draw ROC curve for them for comparison ( with best high resolution )  , you told me befor this is not available in knowledge flow , how can i do this using explorer or experminter ?  note : i try normal weka o/p
     it's
         not clear and most of line overlapped . with sympols like + & *
       
        The Explorer uses the same code as the KnowledgeFlow. Yes, you can use the KnowledgeFlow to plot ROC curves. You can also save the data for a plot into a file, using the Save? button in the plotting window. Then you can use an external tool to make a high-resolution
         plot.
       
        > 2-  i worked with vote classifiers vote ensemble in weka spark  , the output model for JRip ( contains 4 model) as i sliced my dataset to 4 partitions ( please see attached file) , but how can i know the best winning model between the 4 models by voting.
       
        Voting will use all the models. It?s does not use a single model for prediction.
       
        > 3-  if i reference my work in research paper for experiments which are done by weka -spark, it is enough these two references or i need more , specially for weka-spark ??
       
        > M.Hall, Mark, Eibe Frank, Geoffrey Holmes, Bernhard Pfahringer, Peter Reutemann, and Ian H. Witten. "The WEKA data mining software: an update." ACM SIGKDD explorations newsletter 11, no. 1 (2009): 10-18.
        >
        > [69] Eibe Frank, Mark A. Hall, and Ian H. Witten (2016). The WEKA Workbench. Online Appendix for "Data Mining: Practical Machine Learning Tools and Techniques", Morgan Kaufmann, Fourth Edition, 2016.
       
        That sounds fine.
       
        Cheers,
        Eibe
       
       
       
        ------------------------------
       
        Message: 3
        Date: Wed, 7 Jun 2017 20:22:59 +1200
        From: Eibe Frank <[hidden email]>
        To: "Weka machine learning workbench list."
                <[hidden email]>
        Subject: Re: [Wekalist] AutoWEKA on Windows
        Message-ID: <[hidden email]>
        Content-Type: text/plain; charset=utf-8
       
        I use WEKA on Windows 10 all the time, and Windows is the most popular platform for running WEKA. There must be something specific to your particular set-up that's causing problems.
       
        Cheers,
        Eibe
       
        > On 6 Jun 2017, at 21:32, Alexander Osherenko <[hidden email]> wrote:
        >
        > The problem in Weka installation on Win10 is RunWeka.bat. If I comment out the first line in it I can see that last line cant't be executed if called on the console from outside the Weka directory (works good if called from the Weka directory):
        >
        > javaw -classpath . RunWeka -i .\RunWeka.ini -w .\weka.jar -c default ""
        >
        > Consequently, although the Weka shortcut specifies the Run in directory, the target ..\Weka-3-8\RunWeka.bat default doesn't work. Specifying instead the target ..\Weka-3-8\debug_runWeka.bat (my own) that contains one line
        >
        > java -jar ...\Weka-3-8\weka.jar
        >
        > shows up Weka started from Win10 menu.
        >
        > Best, Alexander?
        >
        > 2017-06-05 12:43 GMT+01:00 Alexander Osherenko <[hidden email]>:
        > There are problems with paths on Win10.
        >
        > 1. It begins with installing Weka that can't start from menu although the shortcut seems to be OK: Target and Start in folder are correct, but RunWeka.bat doesn't work. Windows console flashes but doesn't start Weka -- I assume, Windows doesn't find RunWeka.bat
         or RunWeka.bat crashes.
        >
        > ?Whatsoever, ??I developed a batch ??workaround ?file that does exactly the same as the shortcut should:
        > set _JAVA_OPTIONS=-Djava.class.path=./weka.jar;
        > cd D:\toDeleteJDK\Weka-3-8
        > d:
        > D:\toDeleteJDK\Weka-3-8\RunWeka.bat default
        >
        > 2. This starts Weka and EVEN the Weka Explorer )). But there are other problems: The AutoWeka package is installed on the hard drive but not in Weka -- AutoWeka classes are not found what I can see on the Weka log, for example, instantiation of AutoWEKA
     classes
         fails. I added autoweka.jar manually to the first line of my batch file.
        >
        > _JAVA_OPTIONS=-Djava.class.path=./weka.jar;C:\Users\osher\wekafiles\packages\Auto-WEKA\autoweka.jar;
        >
        > and now Weka finds necessary AutoWeka classes but the path problem is not solved: I see no instantiation errors on the log anymore, but other error messages (AutoWeka complains about failures), for example,
        > Experiment - Failed to parse trajectory
        >
        > There are no errors on XP at all (neither with the Weka shortcut, nor with the missing autoweka.jar). I assume there are problems with incorrect settings in Win10, but the Weka and the AutoWeka reinstalls to update paths don't help.
        >
        > ?Best, Alexander
        > ?
        > 2017-06-05 5:30 GMT+01:00 Eibe Frank <[hidden email]>:
        > Yes, on my Mac, the following works:
        >
        > export _JAVA_OPTIONS="-Dsun.java.command=dummy -Djava.home=/Library/Java/JavaVirtualMachines/jdk1.8.0_102.jdk/Contents/Home/jre"
        >
        > Cheers,
        > Eibe
        >
        > PS: Interestingly, I had to set -Djava.home to .../jre to make WEKA work, but the /jre part can be omitted when using the JAVA_HOME environment variable instead.
        >
        >
        > > On 5 Jun 2017, at 07:08, Peter Reutemann <[hidden email]> wrote:
        > >
        > > On June 5, 2017 2:57:55 AM GMT+12:00, Alexander Osherenko <[hidden email]> wrote:
        > > Is it possible to set several variables in JAVA_OPTIONS, for example,
        > > to set _JAVA_OPTIONS to sun.java.command=var1 and java.home=var2?
        > >
        > > If I specify string -Dsun.java.command=var1;java.home=var2, it is not parsed correctly in Systeminfo.
        > >
        > > Cheers, Alexander
        > >
        > > 2017-06-03 8:59 GMT+01:00 Alexander Osherenko <[hidden email]>:
        > > Got it, thanx. It is also possible to set the sun.java.command in Systeminfo by setting
        > > _JAVA_OPTIONS=-Dsun.java.command="dummy".
        > >
        > > Best, Alexander
        > >
        > > 2017-06-03 5:44 GM??T+01:00 Eibe Frank <[hidden email]>:
        > >
        > > Yes, modifying RunWeka.bat should work too.
        > >
        > > The advantage of using an environment variable is that you won't have to fix this problem again when you install a different version of WEKA. Also, it works even if you don't have admin rights. A correct setting for JAVA_HOME may also be useful for
     other
         applications, so from that point of view it may also make sense to set this globally in an environment variable.
        > >
        > > The advantage of using _JAVA_OPTIONS in particular is that any settings specified in this variable take precedence over other settings provided for the java command, see
        > >
        > >
       
    https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts> <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts>
     <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts>
        Difference between _JAVA_OPTIONS JAVA ...
         - Stack Overflow <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts>
        stackoverflow.com
        I thought it would be great to have a comparison between _JAVA_OPTIONS and JAVA_TOOL_OPTIONS. I have been searching a bit for one, but I cannot find anything, so I ...
       
       
       
        > >
        > > Cheers,
        > > Eibe
        > >
        > > > On 3 Jun 2017, at 16:08, Alexander Osherenko <[hidden email]> wrote:
        > > >
        > > > I wonder why you are talking about this variable. Is it the only way to set java.home variable in Weka or I can set it in RunWeka.bat?
        > > >
        > > > ?Cheers, Alexander
        > > > ?
        > > > 2017-06-02 23:10 GMT+01:00 Eibe Frank <[hidden email]>:
        > > > Try setting a new variable
        > > >
        > > >   _JAVA_OPTIONS
        > > >
        > > > as a Windows environment variable and give it the value
        > > >
        > > >   -Djava.home=...
        > > >
        > > > Cheers,
        > > > Eibe
        > > >
        > > >
        > > >
        > > > Sent from my iPad
        > > > On 3/06/2017, at 7:15 AM, Alexander Osherenko <[hidden email]> wrote:
        > > >
        > > >> I tried to find the problem and didn't. Maybe you remember I have Weka 3-8-1 and AutoWEKA installed and they are running without any problems on XP. That's why I have to seek further.
        > > >>
        > > >> Let us please compare the settings in Weka Systeminfo. I found some differences, for example, in java.home and sun.java.command. Can I set these variables in RunWeka.bat?
        > > >>
        > > >> 2017-06-02 9:24 GMT+01:00 Michael Hall <[hidden email]>:
        > > >> > On Jun 1, 2017, at 8:48 PM, Eibe Frank <[hidden email]> wrote:
        > > >> >
        > > >> > With Oracle JVMs, you can use the _JAVA_OPTIONS environment variable to pass such command-line parameters to the JVM. This works on all platforms.
        > > >>
        > > >>
        > > >> I wasn?t aware of that one. It should work for this then, I would think. Unless Java itself sets java.home and overrides the one you specify.
        > > >>
        > > >> I didn?t entirely follow this thread and I am not sure why it is necessary. A jvm to run the application is obviously being found and if the same one is used to set the java.home property you would think AutoWeka would work with that? Possibly some
     localization
         issue as the original poster indicated. Then setting the property could work.
        > > >>
        > > >> The other possibility would be that it is the same situation as on OS X. For that the ?java? executable command is not included in the Weka application embedded jre. So AutoWeka can?t find it there and fails to Runtime exec java. I still believe
     this
         is the behavior of the appbundler version that is being used to create the application bundle. Probably just to save some space the executables are omitted. I also still think a fix there would be for whatever is used for the application build process
     to create
         a ?bin' directory in the embedded jre and copy in the ?java? executable command. This did in fact work for me, doing it myself, to get AutoWeka going on OS X.
        > > >>
        > > >> For what this has to do with Windows, if you try to override java.home I would make sure you are pointing it at a jre/jdk that includes bin/java or bin/java.exe.
        > > >>
        > > >> Michael Hall
        > > >>
        > > >>
        > > >> _______________________________________________
        > > >> Wekalist mailing list
        > > >> Send posts to: [hidden email]
        > > >> List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > > >> List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        > > >>
        > > >> _______________________________________________
        > > >> Wekalist mailing list
        > > >> Send posts to: [hidden email]
        > > >> List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > > >> List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        > > >
        > > > _______________________________________________
        > > > Wekalist mailing list
        > > > Send posts to: [hidden email]
        > > > List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > > > List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        > > >
        > > >
        > > > _______________________________________________
        > > > Wekalist mailing list
        > > > Send posts to: [hidden email]
        > > > List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > > > List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        > >
        > > _______________________________________________
        > > Wekalist mailing list
        > > Send posts to: [hidden email]
        > > List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > > List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        > >
        > >
        > >
        > > Separate the options with blanks instead of semicolons. This variable acts like a partial command-line string, which gets added to the actual command-line to start the jvm.
        > >
        > > Cheers, Peter
        > > --
        > > Peter Reutemann
        > > Dept. of Computer Science
        > > University of Waikato, NZ
        > > +64 (7) 858-5174
        > >
        http://www.cms.waikato.ac.nz/~fracpete/ <http://www.cms.waikato.ac.nz/~fracpete/>home | fracpete <http://www.cms.waikato.ac.nz/~fracpete/>
        www.cms.waikato.ac.nz <http://www.cms.waikato.ac.nz>
        Open-source advocate and programmer by heart, specialized in workflow engines, machine learning and data mining.
       
       
       
        > >
        http://www.data-mining.co.nz <http://www.data-mining.co.nz>Data Mining : University of Waikato <http://www.data-mining.co.nz/>
        www.data-mining.co.nz <http://www.data-mining.co.nz>
        The University of Waikato is the home of several open-source software packages for data mining: WEKA - batch learning (preprocessing, classification, regression ...
       
       
       
        > > _______________________________________________
        > > Wekalist mailing list
        > > Send posts to: [hidden email]
        > > List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > > List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        >
        > _______________________________________________
        > Wekalist mailing list
        > Send posts to: [hidden email]
        > List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        >
        >
        > _______________________________________________
        > Wekalist mailing list
        > Send posts to: [hidden email]
        > List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
       
       
       
        ------------------------------
       
        _______________________________________________
        Wekalist mailing list
        [hidden email]
        https://list.waikato.ac.nz/mailman/listinfo/wekalistWekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
       
       
        End of Wekalist Digest, Vol 172, Issue 24
        *****************************************
       
       
       
       
       
        _______________________________________________
        Wekalist mailing list
        Send posts to: [hidden email]
        List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.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/~ml/weka/mailinglist_etiquette.html>
       
   
   
   
   
    ------------------------------
   
    _______________________________________________
    Wekalist mailing list
    [hidden email]
    https://list.waikato.ac.nz/mailman/listinfo/wekalist
   
   
    End of Wekalist Digest, Vol 172, Issue 64
    *****************************************
   
   
   
   
   
    _______________________________________________
    Wekalist mailing list
    Send posts to: [hidden email]
    List info and subscription status: 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]
List info and subscription status: 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: Weka-Spark

MOHAMMED KAMAL
In reply to this post by MOHAMMED KAMAL


I am comparing  F-measure with AUC   not  accuracy  with   AUC   for imbalanced dataset
can i depend f-measure in comparison??  or do you recommend only AUC??






From: [hidden email] <[hidden email]> on behalf of [hidden email] <[hidden email]>
Sent: Monday, June 19, 2017 2:38 AM
To: [hidden email]
Subject: Wekalist Digest, Vol 172, Issue 66
 
Send Wekalist mailing list submissions to
        [hidden email]

To subscribe or unsubscribe via the World Wide Web, visit
        https://list.waikato.ac.nz/mailman/listinfo/wekalist


or, via email, send a message with subject or body 'help' to
        [hidden email]

You can reach the person managing the list at
        [hidden email]

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Wekalist digest..."


Today's Topics:

   1. Re: Weka-Spark (Mark Hall)


----------------------------------------------------------------------

Message: 1
Date: Mon, 19 Jun 2017 11:38:26 +1200
From: Mark Hall <[hidden email]>
To: "Weka machine learning workbench list."
        <[hidden email]>
Subject: Re: [Wekalist] Weka-Spark
Message-ID: <[hidden email]>
Content-Type: text/plain;       charset="UTF-8"

If you are dealing with imbalanced class distributions, then measures such as AUC and AUPRC are more appropriate than accuracy.

You can save the data for each ROC curve in the Explorer from the scatter plot that pops up when you visualize a threshold curve. There are various tools that can handle plotting more than one series of data ? Excel is one; gnuplot is another. It is beyond the scope of this forum to explain how to use such tools.

Cheers,
Mark.

On 19/06/17, 11:20 AM, "ENGMohammed kamal" <[hidden email] on behalf of [hidden email]> wrote:

    Hi Mark,
    yes for final model evaluation
    Mark you comment only on first question
    do you have answers for the next two questions
    cheers
    M.kamal
   
   
    ________________________________________
   
   
   
   
    ________________________________________
    From: [hidden email] <[hidden email]> on behalf of [hidden email] <[hidden email]>
    Sent: Monday, June 19, 2017 1:32 AM
    To: [hidden email]
    Subject: Wekalist Digest, Vol 172, Issue 64
   
   
    Send Wekalist mailing list submissions to
            [hidden email]
   
    To subscribe or unsubscribe via the World Wide Web, visit
           
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>




    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    or, via email, send a message with subject or body 'help' to
            [hidden email]
   
    You can reach the person managing the list at
            [hidden email]
   
    When replying, please edit your Subject line so it is more specific
    than "Re: Contents of Wekalist digest..."
   
   
    Today's Topics:
   
       1. Re: Regression with M5p (Eibe Frank)
       2. Re: Weka-Spark (Mark Hall)
   
   
    ----------------------------------------------------------------------
   
    Message: 1
    Date: Mon, 19 Jun 2017 10:23:19 +1200
    From: Eibe Frank <[hidden email]>
    To: "Weka machine learning workbench list."
            <[hidden email]>
    Subject: Re: [Wekalist] Regression with M5p
    Message-ID: <[hidden email]>
    Content-Type: text/plain; charset=us-ascii
   
    The classifyInstance() method is the correct method to use. Why is a small error problematic?
   
    Cheers,
    Eibe
   
    > On 18/06/2017, at 8:49 PM, Gaetano <[hidden email]> wrote:
    >
    > Hi Eibe
    > I would like to ask you to make numerical prediction for each instance, the
    > only available statement is "classifyInstance" istruction (that i write
    > before)?
    > Because afterwards, when I calculate the rmse, I get very low values
    > (example 2.325554E-5) and I can not understand the reason.
    > Thanks
    >
    >
    >
    > --
    > View this message in context:
    http://weka.8497.n7.nabble.com/Regression-with-M5p-tp40847p40980.html <http://weka.8497.n7.nabble.com/Regression-with-M5p-tp40847p40980.html>WEKA - Regression with M5p <http://weka.8497.n7.nabble.com/Regression-with-M5p-tp40847p40980.html>




    weka.8497.n7.nabble.com
    Regression with M5p. Hi I have a test set and a training set and I should perform a numeric prediction of the class label using M5p. How it is possible to invoke m5p using java weka api? I tried this...
   
   
   
    > Sent from the WEKA mailing list archive at Nabble.com.
    > _______________________________________________
    > Wekalist mailing list
    > Send posts to: [hidden email]
    > List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>




    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
    > List etiquette:
    http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>




    www.cs.waikato.ac.nz


    Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
   
   
   
   
   
   
    ------------------------------
   
    Message: 2
    Date: Mon, 19 Jun 2017 10:32:47 +1200
    From: Mark Hall <[hidden email]>
    To: "Weka machine learning workbench list."
            <[hidden email]>
    Subject: Re: [Wekalist] Weka-Spark
    Message-ID: <[hidden email]>
    Content-Type: text/plain;       charset="UTF-8"
   
    I don't think there is a problem using SMOTE in the evaluation job, as only the partial training folds will be affected by the SMOTE process. What is the purpose of running the classifier job before the evaluation job? Is this just to create a final model?
   
    Cheers,
    Mark.
   
   
    On 19/06/17, 9:18 AM, "ENGMohammed kamal" <[hidden email] on behalf of [hidden email]> wrote:
   
        Hi Eibe,
        >
        > when i put
         sampling techniques only in weka spark job and remove
         it from weka spark evaluation job (no change in values
         ) .. why ?
       
        Q1. Can you be more specific about what you did?
        __________________________________________
       
       
        i put  the following sequence :
        arff header spark job (4 data slices)  ---> randomly shuffle data spark job  ----->
         i add  sampling technique (SMOTE for example) + weka classifier spark job --------->   weka classifier evaluation without sampling (as you recommensd me before)
       
        nothing change in accuracy and f-measure ??
        if i put sampling also in weka classification evaluation spark job i get higher accuracy and f-measure
   
        note: you recommend me to do not that
        ________________________________________
        Q2. if i compare 10 classifiers without sampling and with sampling techniques (class balancer-ROS-RUS-Smote)  on imbalanced dataset what measure should i put in the table for comparison ?
        is it enough to put accuracy , average f-measure and f measure for minority class ? what kind of graphics will be demonstrative for academic papers ?
   
        ------------------------------------------------------------------------------
        Q3, i asked about  ROC curve to compare these classifiers
        your reply : The Explorer uses the same code as the KnowledgeFlow.
         Yes, you can use the KnowledgeFlow to plot ROC curves. You can also save the data for a plot into a file, using the Save? button in the plotting window. Then you can use an external tool to make a high-resolution plot.
        please can you give me further information, i didn't understand this point
        do you mean to make a plot for every classifier and save it separately and then aggregate all with another tool?   which tool do you recommend ?   how can i do this?
   
        i have 10 classifiers ( without - with CB-RUS-ROS-SMOTE)
        50 readings , so will be huge to compare figures
        i think i can give a figure compare each classifier with sampling with / without sampling techniques using ROC curve?
   
        or do you recommend to make bar chart for F-Measure in minority class only
        i want your recommendation
       
       
        cheers
        M.kamal
       
       
       
       
       
       
       
       
       
       
        ________________________________________
        From: [hidden email] <[hidden email]> on behalf of [hidden email] <[hidden email]>
        Sent: Wednesday, June 7, 2017 11:23 AM
        To: [hidden email]
        Subject: Wekalist Digest, Vol 172, Issue 24
       
       
        Send Wekalist mailing list submissions to
                [hidden email]
       
        To subscribe or unsubscribe via the World Wide Web, visit
               
       
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist> <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info Page - University of Waikato
     <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
    Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
    Wekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
    list.waikato.ac.nz
    The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
   
   
   
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        or, via email, send a message with subject or body 'help' to
                [hidden email]
       
        You can reach the person managing the list at
                [hidden email]
       
        When replying, please edit your Subject line so it is more specific
        than "Re: Contents of Wekalist digest..."
       
       
        Today's Topics:
       
           1. Re: LSA implementation (Andria Lan)
           2. Re: Weka-Spark (Eibe Frank)
           3. Re: AutoWEKA on Windows (Eibe Frank)
       
       
        ----------------------------------------------------------------------
       
        Message: 1
        Date: Wed, 7 Jun 2017 09:36:59 +0800
        From: Andria Lan <[hidden email]>
        To: "Weka machine learning workbench list."
                <[hidden email]>
        Subject: Re: [Wekalist] LSA implementation
        Message-ID:
                <CAHM6HB0gqV2YGySomFSd0ToMObVm0Vg7KXDWEaM-W7vhekS=[hidden email]>
        Content-Type: text/plain; charset="utf-8"
       
        Many thanks, Eibe. Highly appreciat your extensive help.
       
        Good luck to you.
       
        Kind regards,
        Andria
        On 6 Jun 2017 6:48 a.m., "Eibe Frank" <[hidden email]> wrote:
       
        >
        > > On 5/06/2017, at 10:27 PM, Andria Lan <[hidden email]> wrote:
        > >
        > > Thanks a lot for the extensive information. Finally, just wanted to
        > confirm the following issue:
        > >
        > > > GridSearch only works for exactly two parameters, and
        > CVParameterSelection cannot optimise nested parameters.
        > >
        > > Did you mean that MultiSearch works similar to GridSearch for the goal
        > of parameters optimisation. Right?
        >
        > Yes. GridSearch has the option to extend the grid if necessary, which
        > MultiSearch doesn?t have, but it can only deal with optimisation problems
        > involving exactly two parameters.
        >
        > Cheers,
        > Eibe
        >
        >
        > _______________________________________________
        > Wekalist mailing list
        > Send posts to: [hidden email]
        > List info and subscription status:
       
    https://list.waikato.ac.nz/ <https://list.waikato.ac.nz/> <https://list.waikato.ac.nz/>Mailing Lists at The University of Waikato <https://list.waikato.ac.nz/>
    Mailing Lists at The University of Waikato <https://list.waikato.ac.nz/>
    list.waikato.ac.nz
    University Mailing Lists. ITS provides mailing list services for all Waikato University departments that want to use them. Lists can be maintained manually, or ...
   
   
   
        list.waikato.ac.nz
        University Mailing Lists. ITS provides mailing list services for all Waikato University departments that want to use them. Lists can be maintained manually, or ...
       
       
       
        > mailman/listinfo/wekalist
        > List etiquette:
        http://www.cs.waikato.ac.nz/~ <http://www.cs.waikato.ac.nz/~>
        > ml/weka/mailinglist_etiquette.html
        >
        -------------- next part --------------
        An HTML attachment was scrubbed...
        URL: <http://list.waikato.ac.nz/pipermail/wekalist/attachments/20170607/0a9417a3/attachment-0001.html>
       
        ------------------------------
       
        Message: 2
        Date: Wed, 7 Jun 2017 16:47:54 +1200
        From: Eibe Frank <[hidden email]>
        To: "Weka machine learning workbench list."
                <[hidden email]>
        Subject: Re: [Wekalist] Weka-Spark
        Message-ID: <[hidden email]>
        Content-Type: text/plain; charset=utf-8
       
       
        > On 7/06/2017, at 8:34 AM, ENGMohammed kamal <[hidden email]> wrote:
        >
        > when i put sampling techniques only in weka spark job and remove it from weka spark evaluation job (no change in values ) .. why ?
       
        Can you be more specific about what you did?
       
        > i have more 3 questions
        > 1- if i have 10 classifiers and iwant to draw ROC curve for them for comparison ( with best high resolution )  , you told me befor this is not available in knowledge flow , how can i do this using explorer or experminter ?  note : i try normal weka o/p
     it's
         not clear and most of line overlapped . with sympols like + & *
       
        The Explorer uses the same code as the KnowledgeFlow. Yes, you can use the KnowledgeFlow to plot ROC curves. You can also save the data for a plot into a file, using the Save? button in the plotting window. Then you can use an external tool to make a high-resolution
         plot.
       
        > 2-  i worked with vote classifiers vote ensemble in weka spark  , the output model for JRip ( contains 4 model) as i sliced my dataset to 4 partitions ( please see attached file) , but how can i know the best winning model between the 4 models by voting.
       
        Voting will use all the models. It?s does not use a single model for prediction.
       
        > 3-  if i reference my work in research paper for experiments which are done by weka -spark, it is enough these two references or i need more , specially for weka-spark ??
       
        > M.Hall, Mark, Eibe Frank, Geoffrey Holmes, Bernhard Pfahringer, Peter Reutemann, and Ian H. Witten. "The WEKA data mining software: an update." ACM SIGKDD explorations newsletter 11, no. 1 (2009): 10-18.
        >
        > [69] Eibe Frank, Mark A. Hall, and Ian H. Witten (2016). The WEKA Workbench. Online Appendix for "Data Mining: Practical Machine Learning Tools and Techniques", Morgan Kaufmann, Fourth Edition, 2016.
       
        That sounds fine.
       
        Cheers,
        Eibe
       
       
       
        ------------------------------
       
        Message: 3
        Date: Wed, 7 Jun 2017 20:22:59 +1200
        From: Eibe Frank <[hidden email]>
        To: "Weka machine learning workbench list."
                <[hidden email]>
        Subject: Re: [Wekalist] AutoWEKA on Windows
        Message-ID: <[hidden email]>
        Content-Type: text/plain; charset=utf-8
       
        I use WEKA on Windows 10 all the time, and Windows is the most popular platform for running WEKA. There must be something specific to your particular set-up that's causing problems.
       
        Cheers,
        Eibe
       
        > On 6 Jun 2017, at 21:32, Alexander Osherenko <[hidden email]> wrote:
        >
        > The problem in Weka installation on Win10 is RunWeka.bat. If I comment out the first line in it I can see that last line cant't be executed if called on the console from outside the Weka directory (works good if called from the Weka directory):
        >
        > javaw -classpath . RunWeka -i .\RunWeka.ini -w .\weka.jar -c default ""
        >
        > Consequently, although the Weka shortcut specifies the Run in directory, the target ..\Weka-3-8\RunWeka.bat default doesn't work. Specifying instead the target ..\Weka-3-8\debug_runWeka.bat (my own) that contains one line
        >
        > java -jar ...\Weka-3-8\weka.jar
        >
        > shows up Weka started from Win10 menu.
        >
        > Best, Alexander?
        >
        > 2017-06-05 12:43 GMT+01:00 Alexander Osherenko <[hidden email]>:
        > There are problems with paths on Win10.
        >
        > 1. It begins with installing Weka that can't start from menu although the shortcut seems to be OK: Target and Start in folder are correct, but RunWeka.bat doesn't work. Windows console flashes but doesn't start Weka -- I assume, Windows doesn't find RunWeka.bat
         or RunWeka.bat crashes.
        >
        > ?Whatsoever, ??I developed a batch ??workaround ?file that does exactly the same as the shortcut should:
        > set _JAVA_OPTIONS=-Djava.class.path=./weka.jar;
        > cd D:\toDeleteJDK\Weka-3-8
        > d:
        > D:\toDeleteJDK\Weka-3-8\RunWeka.bat default
        >
        > 2. This starts Weka and EVEN the Weka Explorer )). But there are other problems: The AutoWeka package is installed on the hard drive but not in Weka -- AutoWeka classes are not found what I can see on the Weka log, for example, instantiation of AutoWEKA
     classes
         fails. I added autoweka.jar manually to the first line of my batch file.
        >
        > _JAVA_OPTIONS=-Djava.class.path=./weka.jar;C:\Users\osher\wekafiles\packages\Auto-WEKA\autoweka.jar;
        >
        > and now Weka finds necessary AutoWeka classes but the path problem is not solved: I see no instantiation errors on the log anymore, but other error messages (AutoWeka complains about failures), for example,
        > Experiment - Failed to parse trajectory
        >
        > There are no errors on XP at all (neither with the Weka shortcut, nor with the missing autoweka.jar). I assume there are problems with incorrect settings in Win10, but the Weka and the AutoWeka reinstalls to update paths don't help.
        >
        > ?Best, Alexander
        > ?
        > 2017-06-05 5:30 GMT+01:00 Eibe Frank <[hidden email]>:
        > Yes, on my Mac, the following works:
        >
        > export _JAVA_OPTIONS="-Dsun.java.command=dummy -Djava.home=/Library/Java/JavaVirtualMachines/jdk1.8.0_102.jdk/Contents/Home/jre"
        >
        > Cheers,
        > Eibe
        >
        > PS: Interestingly, I had to set -Djava.home to .../jre to make WEKA work, but the /jre part can be omitted when using the JAVA_HOME environment variable instead.
        >
        >
        > > On 5 Jun 2017, at 07:08, Peter Reutemann <[hidden email]> wrote:
        > >
        > > On June 5, 2017 2:57:55 AM GMT+12:00, Alexander Osherenko <[hidden email]> wrote:
        > > Is it possible to set several variables in JAVA_OPTIONS, for example,
        > > to set _JAVA_OPTIONS to sun.java.command=var1 and java.home=var2?
        > >
        > > If I specify string -Dsun.java.command=var1;java.home=var2, it is not parsed correctly in Systeminfo.
        > >
        > > Cheers, Alexander
        > >
        > > 2017-06-03 8:59 GMT+01:00 Alexander Osherenko <[hidden email]>:
        > > Got it, thanx. It is also possible to set the sun.java.command in Systeminfo by setting
        > > _JAVA_OPTIONS=-Dsun.java.command="dummy".
        > >
        > > Best, Alexander
        > >
        > > 2017-06-03 5:44 GM??T+01:00 Eibe Frank <[hidden email]>:
        > >
        > > Yes, modifying RunWeka.bat should work too.
        > >
        > > The advantage of using an environment variable is that you won't have to fix this problem again when you install a different version of WEKA. Also, it works even if you don't have admin rights. A correct setting for JAVA_HOME may also be useful for
     other
         applications, so from that point of view it may also make sense to set this globally in an environment variable.
        > >
        > > The advantage of using _JAVA_OPTIONS in particular is that any settings specified in this variable take precedence over other settings provided for the java command, see
        > >
        > >
       
    https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts> <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts>
     <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts>
        Difference between _JAVA_OPTIONS JAVA ...
         - Stack Overflow <https://stackoverflow.com/questions/28327620/difference-between-java-options-java-tool-options-and-java-opts>
        stackoverflow.com
        I thought it would be great to have a comparison between _JAVA_OPTIONS and JAVA_TOOL_OPTIONS. I have been searching a bit for one, but I cannot find anything, so I ...
       
       
       
        > >
        > > Cheers,
        > > Eibe
        > >
        > > > On 3 Jun 2017, at 16:08, Alexander Osherenko <[hidden email]> wrote:
        > > >
        > > > I wonder why you are talking about this variable. Is it the only way to set java.home variable in Weka or I can set it in RunWeka.bat?
        > > >
        > > > ?Cheers, Alexander
        > > > ?
        > > > 2017-06-02 23:10 GMT+01:00 Eibe Frank <[hidden email]>:
        > > > Try setting a new variable
        > > >
        > > >   _JAVA_OPTIONS
        > > >
        > > > as a Windows environment variable and give it the value
        > > >
        > > >   -Djava.home=...
        > > >
        > > > Cheers,
        > > > Eibe
        > > >
        > > >
        > > >
        > > > Sent from my iPad
        > > > On 3/06/2017, at 7:15 AM, Alexander Osherenko <[hidden email]> wrote:
        > > >
        > > >> I tried to find the problem and didn't. Maybe you remember I have Weka 3-8-1 and AutoWEKA installed and they are running without any problems on XP. That's why I have to seek further.
        > > >>
        > > >> Let us please compare the settings in Weka Systeminfo. I found some differences, for example, in java.home and sun.java.command. Can I set these variables in RunWeka.bat?
        > > >>
        > > >> 2017-06-02 9:24 GMT+01:00 Michael Hall <[hidden email]>:
        > > >> > On Jun 1, 2017, at 8:48 PM, Eibe Frank <[hidden email]> wrote:
        > > >> >
        > > >> > With Oracle JVMs, you can use the _JAVA_OPTIONS environment variable to pass such command-line parameters to the JVM. This works on all platforms.
        > > >>
        > > >>
        > > >> I wasn?t aware of that one. It should work for this then, I would think. Unless Java itself sets java.home and overrides the one you specify.
        > > >>
        > > >> I didn?t entirely follow this thread and I am not sure why it is necessary. A jvm to run the application is obviously being found and if the same one is used to set the java.home property you would think AutoWeka would work with that? Possibly some
     localization
         issue as the original poster indicated. Then setting the property could work.
        > > >>
        > > >> The other possibility would be that it is the same situation as on OS X. For that the ?java? executable command is not included in the Weka application embedded jre. So AutoWeka can?t find it there and fails to Runtime exec java. I still believe
     this
         is the behavior of the appbundler version that is being used to create the application bundle. Probably just to save some space the executables are omitted. I also still think a fix there would be for whatever is used for the application build process
     to create
         a ?bin' directory in the embedded jre and copy in the ?java? executable command. This did in fact work for me, doing it myself, to get AutoWeka going on OS X.
        > > >>
        > > >> For what this has to do with Windows, if you try to override java.home I would make sure you are pointing it at a jre/jdk that includes bin/java or bin/java.exe.
        > > >>
        > > >> Michael Hall
        > > >>
        > > >>
        > > >> _______________________________________________
        > > >> Wekalist mailing list
        > > >> Send posts to: [hidden email]
        > > >> List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > > >> List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        > > >>
        > > >> _______________________________________________
        > > >> Wekalist mailing list
        > > >> Send posts to: [hidden email]
        > > >> List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > > >> List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        > > >
        > > > _______________________________________________
        > > > Wekalist mailing list
        > > > Send posts to: [hidden email]
        > > > List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > > > List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        > > >
        > > >
        > > > _______________________________________________
        > > > Wekalist mailing list
        > > > Send posts to: [hidden email]
        > > > List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > > > List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        > >
        > > _______________________________________________
        > > Wekalist mailing list
        > > Send posts to: [hidden email]
        > > List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > > List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        > >
        > >
        > >
        > > Separate the options with blanks instead of semicolons. This variable acts like a partial command-line string, which gets added to the actual command-line to start the jvm.
        > >
        > > Cheers, Peter
        > > --
        > > Peter Reutemann
        > > Dept. of Computer Science
        > > University of Waikato, NZ
        > > +64 (7) 858-5174
        > >
        http://www.cms.waikato.ac.nz/~fracpete/ <http://www.cms.waikato.ac.nz/~fracpete/>home | fracpete <http://www.cms.waikato.ac.nz/~fracpete/>
        www.cms.waikato.ac.nz <http://www.cms.waikato.ac.nz>
        Open-source advocate and programmer by heart, specialized in workflow engines, machine learning and data mining.
       
       
       
        > >
        http://www.data-mining.co.nz <http://www.data-mining.co.nz>Data Mining : University of Waikato <http://www.data-mining.co.nz/>
        www.data-mining.co.nz <http://www.data-mining.co.nz>
        The University of Waikato is the home of several open-source software packages for data mining: WEKA - batch learning (preprocessing, classification, regression ...
       
       
       
        > > _______________________________________________
        > > Wekalist mailing list
        > > Send posts to: [hidden email]
        > > List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > > List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        >
        > _______________________________________________
        > Wekalist mailing list
        > Send posts to: [hidden email]
        > List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
        >
        >
        > _______________________________________________
        > Wekalist mailing list
        > Send posts to: [hidden email]
        > List info and subscription status:
        https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.waikato.ac.nz/mailman/listinfo/wekalist>Wekalist Info
     Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
        > List etiquette:
        http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>Weka
     Mailing List Etiquette - University of Waikato <http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html>
        www.cs.waikato.ac.nz <http://www.cs.waikato.ac.nz>
        Please do NOT send conference announcements, call for papers, etc., to the mailing list. Please do NOT use the mailing list to advertise your publications.
       
       
       
       
       
       
        ------------------------------
       
        _______________________________________________
        Wekalist mailing list
        [hidden email]
        https://list.waikato.ac.nz/mailman/listinfo/wekalistWekalist Info Page - University of Waikato <https://list.waikato.ac.nz/mailman/listinfo/wekalist>
        list.waikato.ac.nz
        The Weka mailing list ([hidden email]) is for discussions pertaining to the use of the Weka machine learning workbench. Weka is a collection of machine ...
       
       
       
       
       
        End of Wekalist Digest, Vol 172, Issue 24
        *****************************************
       
       
       
       
       
        _______________________________________________
        Wekalist mailing list
        Send posts to: [hidden email]
        List info and subscription status:
    https://list.waikato.ac.nz/mailman/listinfo/wekalist <https://list.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/~ml/weka/mailinglist_etiquette.html>
       
   
   
   
   
    ------------------------------
   
    _______________________________________________
    Wekalist mailing list
    [hidden email]
    https://list.waikato.ac.nz/mailman/listinfo/wekalist
   
   
    End of Wekalist Digest, Vol 172, Issue 64
    *****************************************
   
   
   
   
   
    _______________________________________________
    Wekalist mailing list
    Send posts to: [hidden email]
    List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
    List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
   




------------------------------

_______________________________________________
Wekalist mailing list
[hidden email]
https://list.waikato.ac.nz/mailman/listinfo/wekalist


End of Wekalist Digest, Vol 172, Issue 66
*****************************************

_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status: https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html