Time series: Comparing algorithms

--===============7684854166823222508==
Content-Type: multipart/alternative; boundary="0000000000003f431d061c63766e"

--0000000000003f431d061c63766e
Content-Type: text/plain; charset="UTF-8"

Dear all,

I need help in comparing multiple algorithms in a time series application.
Precisely, assuming the number of time units to forecast is 4. In this
case, I will get,.4 Mean absolute error values associated with each
forecasted point (i.e., k-step-ahead). Assuming that I have 2 algorithms to
compare. In this case, 4 Mean absolute error values are generated for each
algorithm. Thus, it is very challenging to find the best algorithm since
there are 4 Mean absolute error values for each algorithm rather than
having one Mean absolute error value to reflect the overall performance
related to the 4 forecasts.

Is it a correct approach to compute the Average of these Mean absolute
error values (I do that myself) for each algorithm and compare the
resulting Average Mean absolute error values of both algorithms to make a
final reasonable conclusion about their overall performance as a step
toward finding the best algorithm? Or is there any way in WEKA to have one
indicative Mean absolute error value for the multiple forecasts of each
algorithm?

I hope I was clear enough. Your help is highly appreciated.

Regards,
Peter

--0000000000003f431d061c63766e
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Dear all,=C2=A0<div><br></div><div>I need help in comparin=
g multiple algorithms in a time series application. Precisely, assuming the=
 number of time units to forecast is 4. In this case, I will get,.4 Mean ab=
solute error values associated with each forecasted point (i.e., k-step-ahe=
ad). Assuming that I have 2 algorithms to compare. In this case, 4 Mean abs=
olute error values are generated for each algorithm. Thus, it is very chall=
enging to find the best algorithm since there are 4 Mean absolute error val=
ues for each algorithm rather than having one Mean absolute error value to =
reflect the overall performance related to the 4 forecasts.=C2=A0</div><div=
><br></div><div>Is it a correct approach to compute the Average of these Me=
an absolute error values (I do that myself) for each algorithm and compare =
the resulting Average Mean absolute error values of both algorithms to make=
 a final reasonable conclusion about their overall performance as a step to=
ward finding the best algorithm? Or is there any way in WEKA to have one in=
dicative Mean absolute error value for the multiple forecasts of each algor=
ithm?</div><div><br></div><div>I hope I was clear enough. Your help is high=
ly appreciated.</div><div><br></div><div>Regards,=C2=A0</div><div>Peter=C2=
=A0=C2=A0</div></div>

--0000000000003f431d061c63766e--

--===============7684854166823222508==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Wekalist mailing list -- [email protected]
Send posts to [email protected]
To unsubscribe send an email to [email protected]
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

--===============7684854166823222508==--