Timeseries forecasting

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

Timeseries forecasting

Edward Wiskers

Hi all,

In timeseries forecasting, what are the cases of using the Lag?

Best,
Edward


_______________________________________________
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: Timeseries forecasting

Eibe Frank-2
Administrator
Lag variables allow you to take the past values of attributes into account to predict future values. On the airline data that comes with the WEKA distribution, the Forecast panel uses a lag of 12 by default, so it uses values from up to 12 time steps in the past in the predictor. You can change this in the "Lag creation" tab.

Cheers,
Eibe

> On 27 May 2017, at 20:03, Edward Wiskers <[hidden email]> wrote:
>
> Hi all,
>
> In timeseries forecasting, what are the cases of using the Lag?
>
> Best,
> Edward
>
> _______________________________________________
> 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: Timeseries forecasting

Edward Wiskers

Thank you so much, Eibe. It's very clear now.

However, do you recommend always ticking the "Use custom lag lengths" before firing any classifier?

Edward

On 28 May 2017 1:41 p.m., "Eibe Frank" <[hidden email]> wrote:
Lag variables allow you to take the past values of attributes into account to predict future values. On the airline data that comes with the WEKA distribution, the Forecast panel uses a lag of 12 by default, so it uses values from up to 12 time steps in the past in the predictor. You can change this in the "Lag creation" tab.

Cheers,
Eibe

> On 27 May 2017, at 20:03, Edward Wiskers <[hidden email]> wrote:
>
> Hi all,
>
> In timeseries forecasting, what are the cases of using the Lag?
>
> Best,
> Edward
>
> _______________________________________________
> 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
Reply | Threaded
Open this post in threaded view
|

Re: Timeseries forecasting

Eibe Frank-2
Administrator
You'll probably have to experiment with different lengths to get the best reduction in error.

Cheers,
Eibe

> On 28 May 2017, at 18:21, Edward Wiskers <[hidden email]> wrote:
>
> Thank you so much, Eibe. It's very clear now.
>
> However, do you recommend always ticking the "Use custom lag lengths" before firing any classifier?
>
> Edward
>
> On 28 May 2017 1:41 p.m., "Eibe Frank" <[hidden email]> wrote:
> Lag variables allow you to take the past values of attributes into account to predict future values. On the airline data that comes with the WEKA distribution, the Forecast panel uses a lag of 12 by default, so it uses values from up to 12 time steps in the past in the predictor. You can change this in the "Lag creation" tab.
>
> Cheers,
> Eibe
>
> > On 27 May 2017, at 20:03, Edward Wiskers <[hidden email]> wrote:
> >
> > Hi all,
> >
> > In timeseries forecasting, what are the cases of using the Lag?
> >
> > Best,
> > Edward
> >
> > _______________________________________________
> > 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
Reply | Threaded
Open this post in threaded view
|

Re: Timeseries forecasting

Edward Wiskers

Thanks a lot for the prompt reply. I'll consider your advice, Eibe.

Edward

On 28 May 2017 2:47 p.m., "Eibe Frank" <[hidden email]> wrote:
You'll probably have to experiment with different lengths to get the best reduction in error.

Cheers,
Eibe

> On 28 May 2017, at 18:21, Edward Wiskers <[hidden email]> wrote:
>
> Thank you so much, Eibe. It's very clear now.
>
> However, do you recommend always ticking the "Use custom lag lengths" before firing any classifier?
>
> Edward
>
> On 28 May 2017 1:41 p.m., "Eibe Frank" <[hidden email]> wrote:
> Lag variables allow you to take the past values of attributes into account to predict future values. On the airline data that comes with the WEKA distribution, the Forecast panel uses a lag of 12 by default, so it uses values from up to 12 time steps in the past in the predictor. You can change this in the "Lag creation" tab.
>
> Cheers,
> Eibe
>
> > On 27 May 2017, at 20:03, Edward Wiskers <[hidden email]> wrote:
> >
> > Hi all,
> >
> > In timeseries forecasting, what are the cases of using the Lag?
> >
> > Best,
> > Edward
> >
> > _______________________________________________
> > 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