How to perform forecast on KnowledgeFlow

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

How to perform forecast on KnowledgeFlow

Daniel LIAO
Dear Weka people, 

I have completed MOOC3 1.3 activity with Explorer, and also tried to replicate using KnowledgeFlow without success. 

I tried a few items searched with 'timeseries' or 'forecast', but none of them resemble the functionalities found in Explorer's Forecast panel. 

Is it possible to do timeseries forecast 1.3 activity in KnowledgeFlow (I recorded all the steps using xmind and attached a picture of it)? I didn't find anything about it on weka manual.

Could anyone give me some guidance? 

Thank you in advance,

Sincerely, 

Daniel

_______________________________________________
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

3.1.3 Activity Evaluating on a test-set.png (76K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: How to perform forecast on KnowledgeFlow

Eibe Frank-2
Administrator
The time series forecasting panel is available as a separate "perspective" in the KnowledgeFlow (KF). You may have to change the KF settings so that this perspective is shown as a tab in the GUI once the time series forecasting package has been installed.

To send some data to this "time series forecasting" perspective, place a data source on the canvas in the default "data mining processes" perspective of the KF, right-click on it to pick a dataset, and choose "Send to all perspectives".

Once you have built a forecasting model in the "Time series forecasting" perspective, you can integrate it into a flow in the "data mining processes" perspective using the "TimeSeriesForecasting" component. Just save the model in the "time series forecasting" perspective and load it into the component in the "data mining processes" perspective.

Cheers,
Eibe

> On 27 Nov 2016, at 01:05, Daniel LIAO <[hidden email]> wrote:
>
> Dear Weka people,
>
> I have completed MOOC3 1.3 activity with Explorer, and also tried to replicate using KnowledgeFlow without success.
>
> I tried a few items searched with 'timeseries' or 'forecast', but none of them resemble the functionalities found in Explorer's Forecast panel.
>
> Is it possible to do timeseries forecast 1.3 activity in KnowledgeFlow (I recorded all the steps using xmind and attached a picture of it)? I didn't find anything about it on weka manual.
>
> Could anyone give me some guidance?
>
> Thank you in advance,
>
> Sincerely,
>
> Daniel
> <3.1.3 Activity  Evaluating on a test-set.png>_______________________________________________
> 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: How to perform forecast on KnowledgeFlow

Daniel LIAO
In reply to this post by Daniel LIAO

Hi Eibe, 

Thank you so much for your help. 

I managed to do the following: 

1. make time series forecasting perspective visible/available by changing the setting inside Data Mining Processes, after installing the time series forecast library; 

2. inside Data Mining Processes, use ArffLoader and send dataset to all perspective, then time series forecasting perspective is available for clicking; 

3. build a model inside time series forecasting perspective just like doing it as in Explorer, then save the model

4. inside Data Mining Processes, use TimeSeriesForecasting to load the model. 

I can make a dataset link between ArffLoader and TimeSeriesForecasting, or between classAssiger and TimeSeriesForecasting; but from TimeSeriesForecasting, there is only instance link available, and nothing else seems to be able to connected with. 

The problem now is I don't know how to make use of it from here onward. 

Thanks a lot,

Sincerely, 

Daniel
 
Message: 1
Date: Sun, 27 Nov 2016 15:21:27 +1300
From: Eibe Frank <[hidden email]>
To: "Weka machine learning workbench list."
        <[hidden email]>
Subject: Re: [Wekalist] How to perform forecast on KnowledgeFlow
Message-ID: <[hidden email]>
Content-Type: text/plain; charset=us-ascii

The time series forecasting panel is available as a separate "perspective" in the KnowledgeFlow (KF). You may have to change the KF settings so that this perspective is shown as a tab in the GUI once the time series forecasting package has been installed.

To send some data to this "time series forecasting" perspective, place a data source on the canvas in the default "data mining processes" perspective of the KF, right-click on it to pick a dataset, and choose "Send to all perspectives".

Once you have built a forecasting model in the "Time series forecasting" perspective, you can integrate it into a flow in the "data mining processes" perspective using the "TimeSeriesForecasting" component. Just save the model in the "time series forecasting" perspective and load it into the component in the "data mining processes" perspective.

Cheers,
Eibe

> On 27 Nov 2016, at 01:05, Daniel LIAO <[hidden email]> wrote:
>
> Dear Weka people,
>
> I have completed MOOC3 1.3 activity with Explorer, and also tried to replicate using KnowledgeFlow without success.
>
> I tried a few items searched with 'timeseries' or 'forecast', but none of them resemble the functionalities found in Explorer's Forecast panel.
>
> Is it possible to do timeseries forecast 1.3 activity in KnowledgeFlow (I recorded all the steps using xmind and attached a picture of it)? I didn't find anything about it on weka manual.
>
> Could anyone give me some guidance?
>
> Thank you in advance,
>
> Sincerely,
>
> Daniel
> <3.1.3 Activity  Evaluating on a test-set.png>_______________________________________________
> 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: Sun, 27 Nov 2016 15:52:19 +1300
From: Eibe Frank <[hidden email]>
To: "Weka machine learning workbench list."
        <[hidden email]>
Subject: Re: [Wekalist] Explorer and KnowledgeFlow produce different
        result
Message-ID: <[hidden email]>
Content-Type: text/plain; charset=us-ascii

You have *multiple* batchClassifier connections between the SMO component and the ClassifierPerformanceEvaluator component. Right-click on the arrow in the flow to see them all. Delete two of the three you have. Then you will get

  38.7908

RMSE in the cross-validation.

You will get the same RMSE in the Explorer if you turn normalization/standardization off in SMOreg. (Note that you have also turned this off in the SMOreg component in your knowledge flow!)

Cheers,
Eibe

> On 27 Nov 2016, at 02:34, Daniel LIAO <[hidden email]> wrote:
>
> Hi Eibe,
>
> Thank you very much for your response and your verification.
>
> However, I tried again on both stable3-8-1 and developer-branch shortsnap, and still got the same results as below:
>
> stable3-8-1: 66%split, Explorer: RMSE 31.54 vs KF: RMSE 32.04
> developer-branch: 66%split, Explorer: RMSE 31.54 vs KF: RMSE 32.04
>
> stable3-8-1: 10-cv, Explorer: RMSE 36.39 vs KF: RMSE 45.71
> developer-branch: 10-cv, Explorer: RMSE 36.39 vs KF: RMSE 45.71
>
>  I have saved an image and the knowledgeFlow file in a google drive link, could you have a look of my KF file?
> I wonder whether it could be some error in my way of doing KF here?
>
> Thanks again
>
> Sincerely,
>
> Daniel
> Message: 1
> Date: Sat, 26 Nov 2016 19:53:44 +1300
> From: Eibe Frank <[hidden email]>
> To: "Weka machine learning workbench list."
>         <[hidden email]>
> Subject: Re: [Wekalist] Explorer and KnowledgeFlow produce different
>         result  on MOOC3, 1.2 Lesson Activity 4
> Message-ID: <[hidden email]>
> Content-Type: text/plain; charset=us-ascii
>
> I get consistent results when using the latest snapshot of WEKA 3.8. These results are somewhat different from what's stated in the MOOC though. I get 36.3312.
>
> Cheers,
> Eibe
>
> > On 26 Nov 2016, at 17:17, Daniel LIAO <[hidden email]> wrote:
> >
> > Dear Weka people,
> >
> > I intend to replicate as much as possible on KnowledgeFlow from Explorer.
> >
> > On MOOC3, Lesson 1.2 Activity 4.  Using Explorer, I can get the correct answer 38.8; but only get 45.7 using KnowledgeFlow (both using 10 fold cv). When I use train/test split by 66%, Explorer and KnowledgeFlow get the same answer on 32.
> >
> > I did make sure right class attribute is selected and the filter dataset in KnowledgeFlow is the same as it is in Explorer.
> >
> > Is it that I still got something wrong?
> >
> > I attached my KF file, could you have a look?
> >
> > Thanks in advance
> >
> > Sinerely
> >
> > Daniel
> > <3.1.2activity4.kf>_______________________________________________
> > 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: Sat, 26 Nov 2016 20:12:10 +1300
> From: Eibe Frank <[hidden email]>
> To: "Weka machine learning workbench list."
>         <[hidden email]>
> Subject: Re: [Wekalist] Attribute Selection for Time Series
> Message-ID: <[hidden email]>
> Content-Type: text/plain; charset=us-ascii
>
> Can't you just use the AttributeSelectedClassifier as the base classifier in the forecasting panel?
>
>
> Cheers,
> Eibe
>
> > On 25 Nov 2016, at 08:50, Nicolae Righeriu <[hidden email]> wrote:
> >
> > Hello,
> > as my bachelor thesis subject in computer science I want to evaluate different feature/attribute selection methods for time series, specifically for load forecasting. As I've used already methods like WrapperSubsetEval, Cfssubset or others I noticed that for univariate time series they have no effect (they see only 2 attributes: date and load); that is because they can work only on the Lag matrix, to see which points in the past are most relevant for the next points. I then built lag matrix and it did choose the most important lags.
> >     However the lag matrix is also built within the forecaster and it would be redundant to build it 2 times, so here is my question: can I somehow use the Attribute selection methods within the created lag? Or can I create my own lag matrix and suppress the lagmaker in the forecaster?
> >     Also my thesis aim isn't yet clearly defined and my assistant said I can for example implement a new, useful search method along the existent GreedyStepWise, BestFirst etc. Do you happen to know what could be a good contribution?
> > _______________________________________________
> > 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: 3
> Date: Sat, 26 Nov 2016 20:13:52 +1300
> From: Mark Hall <[hidden email]>
> To: "Weka machine learning workbench list."
>         <[hidden email]>
> Subject: Re: [Wekalist] Explorer and KnowledgeFlow produce different
>         result on MOOC3, 1.2 Lesson Activity 4
> Message-ID: <[hidden email]>
> Content-Type: text/plain; charset="utf-8"
>
> Somehow (and for the life of me I don't know how you achieved it :-)) there are three batchClassifier connections between your SMOreg step and the ClassifierPerformanceEvaluator ? this is screwing up the results. The Knowledge Flow does not allow you to make more than one batchClassifier connection between a classifier and the ClassifierPerformanceEvaluator step, so I'd love to know how you did this (short of editing the JSON .kf flow file). When I deleted two of the three batchClassifier connections in your flow I got 38.7908 as the RMSE.
>
>
>
> Cheers,
>
> Mark.
>
>
>
>
>
> From: <[hidden email]> on behalf of Daniel LIAO <[hidden email]>
> Reply-To: "Weka machine learning workbench list." <[hidden email]>
> Date: Saturday, 26 November 2016 at 5:17 PM
> To: "Weka machine learning workbench list." <[hidden email]>
> Subject: [Wekalist] Explorer and KnowledgeFlow produce different result on MOOC3, 1.2 Lesson Activity 4
>
>
>
> Dear Weka people,
>
>
>
> I intend to replicate as much as possible on KnowledgeFlow from Explorer.
>
>
>
> On MOOC3, Lesson 1.2 Activity 4.  Using Explorer, I can get the correct answer 38.8; but only get 45.7 using KnowledgeFlow (both using 10 fold cv). When I use train/test split by 66%, Explorer and KnowledgeFlow get the same answer on 32.
>
>
>
> I did make sure right class attribute is selected and the filter dataset in KnowledgeFlow is the same as it is in Explorer.
>
>
>
> Is it that I still got something wrong?
>
>
>
> I attached my KF file, could you have a look?
>
>
>
> Thanks in advance
>
>
>
> Sinerely
>
>
>
> Daniel
>
> _______________________________________________ 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/20161126/2707488f/attachment-0001.html>
>
> ------------------------------
>
> Message: 4
> Date: Sat, 26 Nov 2016 21:16:54 +1300
> From: Eibe Frank <[hidden email]>
> To: "Weka machine learning workbench list."
>         <[hidden email]>
> Subject: Re: [Wekalist] use Experimenter inside KnowledgeFlow and use
>         API in  simpleCLI
> Message-ID: <[hidden email]>
> Content-Type: text/plain; charset=us-ascii
>
>
> > On 25 Nov 2016, at 04:21, Daniel LIAO <[hidden email]> wrote:
> >
> > First, as Eibe commented that KnowledgeFlow can do what the other interfaces do, I wonder can we use experimenter inside KnowledgeFlow(KF), in other words, can KF perform what Experimenter's functionalities? one of key functions I think is, for example, run 3 different classifiers against 5 datasets. Can KF do it somehow?
>
> The KF doesn't support loops. Take a look at ADAMS (https://adams.cms.waikato.ac.nz/) for a workflow environment hat allows you to specify complex processes with repetition.
>
> > Second, Weka API (Java) has a lot of functionalities, what proportion of functionalities provided by API is not covered by  the weka interfaces l used the most (Explorer, Experimenter, KnowledgeFlow)? What are those functionalities which are not covered by these 3 interfaces? When would those functions become important for machine learning practitioners?
>
> Almost all the functionality is available via at least one of the GUIs. However, the GUIs don't offer the same flexibility as writing actual programs based on the API. Some lower-level functionality (e.g., computing simple density estimators) is not available directly via the GUIs.
>
> > Third, for those interfaces-uncovered functionalities, without using Jython or Python-weka or RWeka, can we access the full functions of Weka API inside simpleCLI?
>
> No. Use Groovy or Jython instead if you want to write scripts that use the API.
>
> Cheers,
> Eibe
>
> ------------------------------
>
> Message: 5
> Date: Sat, 26 Nov 2016 20:05:08 +0800
> From: Daniel LIAO <[hidden email]>
> To: "[hidden email]" <[hidden email]>
> Subject: [Wekalist] How to perform forecast on KnowledgeFlow
> Message-ID:
>         <[hidden email]>
> Content-Type: text/plain; charset="utf-8"
>
> Dear Weka people,
>
> I have completed MOOC3 1.3 activity with Explorer, and also tried to
> replicate using KnowledgeFlow without success.
>
> I tried a few items searched with 'timeseries' or 'forecast', but none of
> them resemble the functionalities found in Explorer's Forecast panel.
>
> Is it possible to do timeseries forecast 1.3 activity in KnowledgeFlow (I
> recorded all the steps using xmind and attached a picture of it)? I didn't
> find anything about it on weka manual.
>
> Could anyone give me some guidance?
>
> Thank you in advance,
>
> Sincerely,
>
> Daniel
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <http://list.waikato.ac.nz/pipermail/wekalist/attachments/20161126/e5ec9109/attachment.html>
> -------------- next part --------------
> A non-text attachment was scrubbed...
> Name: 3.1.3 Activity  Evaluating on a test-set.png
> Type: image/png
> Size: 57048 bytes
> Desc: not available
> URL: <http://list.waikato.ac.nz/pipermail/wekalist/attachments/20161126/e5ec9109/attachment.png>
>
> ------------------------------
>
> _______________________________________________
> Wekalist mailing list
> [hidden email]
> https://list.waikato.ac.nz/mailman/listinfo/wekalist
>
>
> End of Wekalist Digest, Vol 165, Issue 71
> *****************************************
>
> _______________________________________________
> 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: 3
Date: Sun, 27 Nov 2016 15:53:44 +1300
From: Eibe Frank <[hidden email]>
To: "Weka machine learning workbench list."
        <[hidden email]>
Subject: Re: [Wekalist] Explorer and KnowledgeFlow produce different
        result  on MOOC3, 1.2 Lesson Activity 4
Message-ID: <[hidden email]>
Content-Type: text/plain; charset=us-ascii

Double-check your connections. Right-click on the corresponding arrow. Even though only one arrow is shown, there are actually multiple connections. You need to delete all but one of them.

Cheers,
Eibe

> On 27 Nov 2016, at 02:45, Daniel LIAO <[hidden email]> wrote:
>
>
> Hi Mark,
>
> Thank you very much for your reply and help.
>
> I understand it could cause error if I make more than one batchClassifer link from one classifier to more than one ClassifierPerformanceEvaluator.
>
> However, I did not do that in the KF file I attached in the previous email, I have two batchClassifier links for two classifier and two ClassifierPerformanceEvaluator.
>
> Also I have checked again in both stable 3-8-1 and developer-branch with the same file and still got the same result.
>
> I have uploaded a snapshot of my kf file, and the kf file itself in this google drive link  , could you have a look?( in which you can see how exactly I did it)
>
> Thanks again,
>
> Sinerely,
>
> Daniel
>
> Message: 3
> Date: Sat, 26 Nov 2016 20:13:52 +1300
> From: Mark Hall <[hidden email]>
> To: "Weka machine learning workbench list."
>         <[hidden email]>
> Subject: Re: [Wekalist] Explorer and KnowledgeFlow produce different
>         result on MOOC3, 1.2 Lesson Activity 4
> Message-ID: <[hidden email]>
> Content-Type: text/plain; charset="utf-8"
>
> Somehow (and for the life of me I don't know how you achieved it :-)) there are three batchClassifier connections between your SMOreg step and the ClassifierPerformanceEvaluator ? this is screwing up the results. The Knowledge Flow does not allow you to make more than one batchClassifier connection between a classifier and the ClassifierPerformanceEvaluator step, so I'd love to know how you did this (short of editing the JSON .kf flow file). When I deleted two of the three batchClassifier connections in your flow I got 38.7908 as the RMSE.
>
>
>
> Cheers,
>
> Mark.
>
>
>
>
>
> From: <[hidden email]> on behalf of Daniel LIAO <[hidden email]>
> Reply-To: "Weka machine learning workbench list." <[hidden email]>
> Date: Saturday, 26 November 2016 at 5:17 PM
> To: "Weka machine learning workbench list." <[hidden email]>
> Subject: [Wekalist] Explorer and KnowledgeFlow produce different result on MOOC3, 1.2 Lesson Activity 4
>
>
>
> Dear Weka people,
>
>
>
> I intend to replicate as much as possible on KnowledgeFlow from Explorer.
>
>
>
> On MOOC3, Lesson 1.2 Activity 4.  Using Explorer, I can get the correct answer 38.8; but only get 45.7 using KnowledgeFlow (both using 10 fold cv). When I use train/test split by 66%, Explorer and KnowledgeFlow get the same answer on 32.
>
>
>
> I did make sure right class attribute is selected and the filter dataset in KnowledgeFlow is the same as it is in Explorer.
>
>
>
> Is it that I still got something wrong?
>
>
>
> I attached my KF file, could you have a look?
>
>
>
> Thanks in advance
>
>
>
> Sinerely
>
>
>
> Daniel
>
> _______________________________________________ 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/20161126/2707488f/attachment-0001.html>
>
> ------------------------------
>
> Message: 4
> Date: Sat, 26 Nov 2016 21:16:54 +1300
> From: Eibe Frank <[hidden email]>
> To: "Weka machine learning workbench list."
>         <[hidden email]>
> Subject: Re: [Wekalist] use Experimenter inside KnowledgeFlow and use
>         API in  simpleCLI
> Message-ID: <[hidden email]>
> Content-Type: text/plain; charset=us-ascii
>
>
> > On 25 Nov 2016, at 04:21, Daniel LIAO <[hidden email]> wrote:
> >
> > First, as Eibe commented that KnowledgeFlow can do what the other interfaces do, I wonder can we use experimenter inside KnowledgeFlow(KF), in other words, can KF perform what Experimenter's functionalities? one of key functions I think is, for example, run 3 different classifiers against 5 datasets. Can KF do it somehow?
>
> The KF doesn't support loops. Take a look at ADAMS (https://adams.cms.waikato.ac.nz/) for a workflow environment hat allows you to specify complex processes with repetition.
>
> > Second, Weka API (Java) has a lot of functionalities, what proportion of functionalities provided by API is not covered by  the weka interfaces l used the most (Explorer, Experimenter, KnowledgeFlow)? What are those functionalities which are not covered by these 3 interfaces? When would those functions become important for machine learning practitioners?
>
> Almost all the functionality is available via at least one of the GUIs. However, the GUIs don't offer the same flexibility as writing actual programs based on the API. Some lower-level functionality (e.g., computing simple density estimators) is not available directly via the GUIs.
>
> > Third, for those interfaces-uncovered functionalities, without using Jython or Python-weka or RWeka, can we access the full functions of Weka API inside simpleCLI?
>
> No. Use Groovy or Jython instead if you want to write scripts that use the API.
>
> Cheers,
> Eibe
>
> ------------------------------
>
> Message: 5
> Date: Sat, 26 Nov 2016 20:05:08 +0800
> From: Daniel LIAO <[hidden email]>
> To: "[hidden email]" <[hidden email]>
> Subject: [Wekalist] How to perform forecast on KnowledgeFlow
> Message-ID:
>         <[hidden email]>
> Content-Type: text/plain; charset="utf-8"
>
> Dear Weka people,
>
> I have completed MOOC3 1.3 activity with Explorer, and also tried to
> replicate using KnowledgeFlow without success.
>
> I tried a few items searched with 'timeseries' or 'forecast', but none of
> them resemble the functionalities found in Explorer's Forecast panel.
>
> Is it possible to do timeseries forecast 1.3 activity in KnowledgeFlow (I
> recorded all the steps using xmind and attached a picture of it)? I didn't
> find anything about it on weka manual.
>
> Could anyone give me some guidance?
>
> Thank you in advance,
>
> Sincerely,
>
> Daniel
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <http://list.waikato.ac.nz/pipermail/wekalist/attachments/20161126/e5ec9109/attachment.html>
> -------------- next part --------------
> A non-text attachment was scrubbed...
> Name: 3.1.3 Activity  Evaluating on a test-set.png
> Type: image/png
> Size: 57048 bytes
> Desc: not available
> URL: <http://list.waikato.ac.nz/pipermail/wekalist/attachments/20161126/e5ec9109/attachment.png>
>
> ------------------------------
>
> _______________________________________________
> Wekalist mailing list
> [hidden email]
> https://list.waikato.ac.nz/mailman/listinfo/wekalist
>
>
> End of Wekalist Digest, Vol 165, Issue 71
> *****************************************
>
>
> _______________________________________________
> 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 165, Issue 74
*****************************************



_______________________________________________
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: How to perform forecast on KnowledgeFlow

falete
It occurs the same to me. From timeseriesforecasting only instance is
available. How can I connect it to an output element? txt, csv, etc.



--
Sent from: https://weka.8497.n7.nabble.com/
_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: How to perform forecast on KnowledgeFlow

falete
I've solved as the image attached
<https://weka.8497.n7.nabble.com/file/t7313/Capture.png>



--
Sent from: https://weka.8497.n7.nabble.com/
_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html