Strange forecast / prediction with TSMODEL

[Mark Lavender]

I've got some time series data that is confined to the range 0-1.

It runs from 2006 thru Nov 2010.  For the years 2006 and 2007, however, we
only have quarterly values, so we only populated months 3,6,9,12 for those
months.

Here's the odd part.  When we use TSMODEL to give a prediction, we get the
following:

Date Predict
2007-3 843.75
2007-6 0.01
2007-9 0.02
2007-12 0.05
2008-3 0.10
2008-6 0.21
2008-9 0.39
2008-12 0.70
2009-1 1.21
2009-2 2.00
2009-3 3.17
2009-4 4.87
2009-5 7.22
2009-6 10.40
2009-7 14.57
2009-8 19.88
2009-9 26.49
2009-10 34.53
2009-11 44.11
2009-12 55.30

As noted above, the actual values run from 0 to 1.  Somehow it looks like a
very strange value is being forecast for the first period  and then slowly
the model tries to return to that value.

Any ideas?

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Ornelas, Fermin-2]

You cannot estimate a time series model with such a number of records. I built mine with at least 80.

Fermin Ornelas

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Mark Lavender
Sent: Wednesday, January 19, 2011 10:49 AM
To: [hidden email]
Subject: Strange forecast / prediction with TSMODEL

I've got some time series data that is confined to the range 0-1.

It runs from 2006 thru Nov 2010.  For the years 2006 and 2007, however, we
only have quarterly values, so we only populated months 3,6,9,12 for those
months.

Here's the odd part.  When we use TSMODEL to give a prediction, we get the
following:

Date Predict
2007-3 843.75
2007-6 0.01
2007-9 0.02
2007-12 0.05
2008-3 0.10
2008-6 0.21
2008-9 0.39
2008-12 0.70
2009-1 1.21
2009-2 2.00
2009-3 3.17
2009-4 4.87
2009-5 7.22
2009-6 10.40
2009-7 14.57
2009-8 19.88
2009-9 26.49
2009-10 34.53
2009-11 44.11
2009-12 55.30

As noted above, the actual values run from 0 to 1.  Somehow it looks like a
very strange value is being forecast for the first period  and then slowly
the model tries to return to that value.

Any ideas?

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD


NOTICE: This e-mail (and any attachments) may contain PRIVILEGED OR CONFIDENTIAL information and is intended only for the use of the specific individual(s) to whom it is addressed.  It may contain information that is privileged and confidential under state and federal law.  This information may be used or disclosed only in accordance with law, and you may be subject to penalties under law for improper use or further disclosure of the information in this e-mail and its attachments. If you have received this e-mail in error, please immediately notify the person named above by reply e-mail, and then delete the original e-mail.  Thank you.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Jason Burke]

Mark,

You will the same number of observations per year, in your case either four or twelve.

Regards,


Jason


Sent from my BeOS

On 20/01/2011, at 4:55 AM, "Ornelas, Fermin" <[hidden email]> wrote:

> You cannot estimate a time series model with such a number of records. I built mine with at least 80.
>
> Fermin Ornelas
>
> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Mark Lavender
> Sent: Wednesday, January 19, 2011 10:49 AM
> To: [hidden email]
> Subject: Strange forecast / prediction with TSMODEL
>
> I've got some time series data that is confined to the range 0-1.
>
> It runs from 2006 thru Nov 2010.  For the years 2006 and 2007, however, we
> only have quarterly values, so we only populated months 3,6,9,12 for those
> months.
>
> Here's the odd part.  When we use TSMODEL to give a prediction, we get the
> following:
>
> Date Predict
> 2007-3 843.75
> 2007-6 0.01
> 2007-9 0.02
> 2007-12 0.05
> 2008-3 0.10
> 2008-6 0.21
> 2008-9 0.39
> 2008-12 0.70
> 2009-1 1.21
> 2009-2 2.00
> 2009-3 3.17
> 2009-4 4.87
> 2009-5 7.22
> 2009-6 10.40
> 2009-7 14.57
> 2009-8 19.88
> 2009-9 26.49
> 2009-10 34.53
> 2009-11 44.11
> 2009-12 55.30
>
> As noted above, the actual values run from 0 to 1.  Somehow it looks like a
> very strange value is being forecast for the first period  and then slowly
> the model tries to return to that value.
>
> Any ideas?
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>
>
> NOTICE: This e-mail (and any attachments) may contain PRIVILEGED OR CONFIDENTIAL information and is intended only for the use of the specific individual(s) to whom it is addressed.  It may contain information that is privileged and confidential under state and federal law.  This information may be used or disclosed only in accordance with law, and you may be subject to penalties under law for improper use or further disclosure of the information in this e-mail and its attachments. If you have received this e-mail in error, please immediately notify the person named above by reply e-mail, and then delete the original e-mail.  Thank you.
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Ryan]

I haven't spent anytime examining the TSMODEL procedure, so I cannot
speak to that procedure. It may or may not be the optimal approach.
Let me just point out that if you have binary or binomial data from
which you would like to predict future probabilities, you might
consider fitting a *type* of generalized linear model/GEE. Whether
such an approach would be appropriate depends on various aspects to
your data that have yet to be detailed.

Ryan

On Wed, Jan 19, 2011 at 12:48 PM, Mark Lavender <[hidden email]> wrote:

> I've got some time series data that is confined to the range 0-1.
>
> It runs from 2006 thru Nov 2010.  For the years 2006 and 2007, however, we
> only have quarterly values, so we only populated months 3,6,9,12 for those
> months.
>
> Here's the odd part.  When we use TSMODEL to give a prediction, we get the
> following:
>
> Date Predict
> 2007-3 843.75
> 2007-6 0.01
> 2007-9 0.02
> 2007-12 0.05
> 2008-3 0.10
> 2008-6 0.21
> 2008-9 0.39
> 2008-12 0.70
> 2009-1 1.21
> 2009-2 2.00
> 2009-3 3.17
> 2009-4 4.87
> 2009-5 7.22
> 2009-6 10.40
> 2009-7 14.57
> 2009-8 19.88
> 2009-9 26.49
> 2009-10 34.53
> 2009-11 44.11
> 2009-12 55.30
>
> As noted above, the actual values run from 0 to 1.  Somehow it looks like a
> very strange value is being forecast for the first period  and then slowly
> the model tries to return to that value.
>
> Any ideas?
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD