TimeSeries SMOreg same/few target values

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

TimeSeries SMOreg same/few target values

José Heitor
Hi,

Using SMOreg as the time-series base learner fails if the target values in
the training set are the same or if only a few of them are unique.

Linear Regression and Multilayer Perceptron algorithms handle these
scenarios without issues, but I would prefer to use SMOreg as it achieves
better results.

Is there a configuration option or some other recommend process to mitigate
this issue?

Thanks,
Jose



--
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: TimeSeries SMOreg same/few target values

Eibe Frank-2
Administrator
If all the target values in your training data are the same, you might as well use ZeroR.

I don’t understand why whether "only a few of them are unique” is relevant. Is SMOreg running into numerical issue on your data because the data is not “sufficiently continuous”?

Cheers,
Eibe

> On 7/05/2020, at 12:07 AM, José Heitor <[hidden email]> wrote:
>
> Hi,
>
> Using SMOreg as the time-series base learner fails if the target values in
> the training set are the same or if only a few of them are unique.
>
> Linear Regression and Multilayer Perceptron algorithms handle these
> scenarios without issues, but I would prefer to use SMOreg as it achieves
> better results.
>
> Is there a configuration option or some other recommend process to mitigate
> this issue?
>
> Thanks,
> Jose
>
>
>
> --
> 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
_______________________________________________
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: TimeSeries SMOreg same/few target values

José Heitor
Hi Eibe,

For some context: I am scripting the learning and inference process in a
large loop, as I have some 4500 individual time-series predictions to make.
(Kaggle Wallmart exercise for 9-months of weekly predictions for 99
departments in 45 stores)

On datasets with no/missing sales data, I impute zero values, so as to be
able to provide data to the learner within the loop. My hope was that, as
with LinearRegression and MultilayerPerceptron learners, it would just
result in zero-valued predictions, rather than falling over?

What I found with datasets with only a couple of entries near the end of the
training set, is that while trying to do evaluation with a dataset split (in
the Explorer), the learner would only see the same values in the training
portion of the split and would thus fail with the same complaint as above.
This does not cause any issues in the scripted runs though, because there it
does not reserve a portion of the data for evaluation and the full dataset
is available for training.

Cheers,
Jose



--
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