BookmarkSubscribeRSS Feed
🔒 This topic is solved and locked. Need further help from the community? Please sign in and ask a new question.

Hello,

Let us say that we are using forecast server and we have created a specific hierarchy e.g. Whole Country--Region--Prefecture. The hierarchy refers to a product and the lower level of the hierarchy denotes how many units of this product are sold in a specific prefecture that belongs to a specific region of the country. Let us also say that our transactional data are accumulated to weekly data through the automatic feature of forecast server.

The result of the previous example will be to have forecasts for the various levels of the hierarchy in weekly interval. Let us say that we also want ot have the forecasts in daily intervals and in monthly intervals? WHat would be the optimal way of doing that?

One solution that i am thinking is to create two more projects in forecast server and when asked by the software about the accumulation of the transactional data to choose in the one projetc monthly and to the other project daily.

So have three projects in total.

Is this a proper way of doing this or you propose somethignn else?

Thnaks in advance,

Andreas

1 ACCEPTED SOLUTION

Accepted Solutions
udo_sas
SAS Employee

Hello -

You might want to check out the HPFTEMPRECONCILE procedure, which reconciles high-frequency forecasts to the low-frequency forecasts in such a way that the accumulation of the reconciled high-frequency forecasts is equal to the low-frequency forecasts.

For details check out: http://support.sas.com/documentation/cdl/en/hpfug/63959/HTML/default/viewer.htm#hpftemprecon_toc.htm

Note that HPFTEMPRECONCILE is currently not supported in SAS Forecast Studio - so you would need to switch to code for the time being.

In particular I would suggest to check out the SAS Forecast Batch Interface, which is documented in Chapter 11 of the SAS Forecast Server 12.1 Administrator’s Guide.

Thanks!

Udo

View solution in original post

7 REPLIES 7
udo_sas
SAS Employee

Hello -

You might want to check out the HPFTEMPRECONCILE procedure, which reconciles high-frequency forecasts to the low-frequency forecasts in such a way that the accumulation of the reconciled high-frequency forecasts is equal to the low-frequency forecasts.

For details check out: http://support.sas.com/documentation/cdl/en/hpfug/63959/HTML/default/viewer.htm#hpftemprecon_toc.htm

Note that HPFTEMPRECONCILE is currently not supported in SAS Forecast Studio - so you would need to switch to code for the time being.

In particular I would suggest to check out the SAS Forecast Batch Interface, which is documented in Chapter 11 of the SAS Forecast Server 12.1 Administrator’s Guide.

Thanks!

Udo

deleted_user
Not applicable

Hi,

In addition to Udos reply, please also bear in mind that if you use PROC HPFTEMPRECON to do temporal reconciliation, the time intervals need to be nested. So for instance reconciling day to weeks is ok but weeks to months do not work since weeks are not nested within months.

Cheers,

Snurre

andreas_zaras
Pyrite | Level 9

Hello,

Thank you very much for your answers. So the reconciliations of forecasts to higher or lower time intervals are not possible using the GUI of Forecast Studio. In a project SAS cosultants can write code to do that and the client will be able to use this requirement. But let us say that a customer wants to do something on his own in a new task without using code, only point and click. What would you suggest as the best option: the one described in the question with the separate projects or maybe use the Enterprise Guide with the prepare time series task and do the recociliations to different time intervals in all the series of the hierarchy? I know that both pocedures are time consuming and legthy but since there is no Point and Click option which one of the two would you suggest (or something different not mentioned here)?

Thnaks in advance,

Andreas

deleted_user
Not applicable

Hi,

First of all, I don't think the standard tasks in Enterprise Guide would help you with reconciliation as the EG time series task only allows you to move from one frequency to another. Reconciliation requires that you have data for both frequencies you want to reconcile.

If you want to use point-and-click, one option could be to create a parameterized stored process setup in EG and use this to create your Forecast Server projects using the Forecast Server administrative macros followed by a stored process that does the temporal reconciliation of forecasts created in 2 separate Forecast Server projects.

Thanks,

Snurre

andreas_zaras
Pyrite | Level 9

Hello Snurre,

Thanks for your answer. What i asked in my original post is how to move from one time frequency to another (Weekly to Daily and Weekly to Monthly). If i understood well Udo said that this cannot be done through the Forecast Server GUI, only with code (HPFTEMPRECONCILE). So i asked what would be the best way if i did not want to use code. One of the thoughts provided was through the EG prepare time series task.

Thanks again,

Andreas.

deleted_user
Not applicable

Hi Andreas,

I think you can use the tasks in EG for moving between frequencies. However, be aware that creating forecasts at the daily level and then accumulate those forecasts the monthly level will not produce the same forecasts as if you are forecasting at the monthly frequency. In order to have forecasts that are coherent at different frequencies you need some mechanism for reconciling them. This is what HPFTEMPRECON does in the frequency dimension.

Thanks,

Snurre

sas-innovate-2024.png

Join us for SAS Innovate April 16-19 at the Aria in Las Vegas. Bring the team and save big with our group pricing for a limited time only.

Pre-conference courses and tutorials are filling up fast and are always a sellout. Register today to reserve your seat.

 

Register now!

Multiple Linear Regression in SAS

Learn how to run multiple linear regression models with and without interactions, presented by SAS user Alex Chaplin.

Find more tutorials on the SAS Users YouTube channel.

Discussion stats
  • 7 replies
  • 2014 views
  • 4 likes
  • 3 in conversation