Question Re: STATISTICAL FORECASTING planned in (FISCAL WEEKS) values differ from interactive to background in Supply Chain Management Q&A

STATISTICAL FORECASTING planned in (FISCAL WEEKS) values differ from interactive to background

Former Member — Fri, 02 Feb 2018 07:29:45 GMT

Hello Team,

We have an issue when running the statistical forecast for specific product / location and customer code- where the values are not same when run in interactive to that of the same being run in background mode.

Interactive Mode - selection id is Show APO Location and filters are Customer code, product and location ( no of CVC selected = 13)

Forecast profile - Model chosen in 56 ( Auto model 2)

History horizon is 156 Weeks

Forecast horizon is 78 Weeks

Fiscal variant is Fiscal week

Interactive results is saved.

Background job is created with the same selection id and running with the same forecast profile at the aggregation level of Location / Material and Customer code.

Background job results differ from that of the interactive results.

Kindly help us understand what could be the issue as why we are not seeing results as consistent.

Re: STATISTICAL FORECASTING planned in (FISCAL WEEKS) values differ from interactive to background

Former Member — Mon, 05 Feb 2018 07:52:34 GMT

Hi,

Please check "Activity" used in background job is having "always use job profile" enable, if not please check whether you have other assignment in /sapapo/msdp_fcst2.

regards,

Sachin

Re: STATISTICAL FORECASTING planned in (FISCAL WEEKS) values differ from interactive to background

former_member412616 — Mon, 12 Feb 2018 10:24:58 GMT

Hello Team,

We could see that the number of cvc's for the selection are 13.

While the forecast is generated in the background, the stat forecast is started for every single one of combinations. You will find 13 combinations and their calculation in the log.

The statistical forecast is started once for selected level in foreground(for all 13cvc's). The values in the historical input key figure are added up.

Drill down at each Location / Material and Customer code to calculate stat forecast for 13 combinations.

Then, compare the results with results of background.

Kindly refer the point 4 of KBA 1667352 for more details.

Regards,

Naga