Adopt Single Result when Interactively Forecasting is not saving

[christopher_smith1]

Hi everyone,

Running into some strange system behavior when choosing a composite forecast model during interactive forecasting and hoping someone can help.  Thanks in advance for reading and assisting

We are running SCM 7.02 patch lvl 6

Here is our scenario and the issue found by end users:

• In lieu of using Automodel2, we wanted to use a 'pick best' forecast model and thus we've created a composite forecasting model in /sapapo/mc96b which bundles 4 univariant forecast models and uses the RMSE error measure as the means for selecting the best forecast model.  This composite model's master profile uses 36 periods of history and 18 periods of forecast, periods are defined as months, and lifecycle planning is active.  The composite model it's self groups 4 separate univariate profiles, each using different forecast strategies (21, 35, 41, 94).  The weighting between all four univariants are equal (100), and the mode is set to automatically select the profile with the smallest RMSE.

• We run this composite model in batch mode as one of our stat models during our monthly stat. forecast run and it works great so I know that its definition is correct and the settings are appropriate.

Here's where the issue come in:

There are cases where the RMSE error reduction pick best design does not yield the 'best' forecast and a planner will want to override what the system selected in the batch run by interactively forecasting the CVC.

• The planner pulls up the CVC in the shuffler and loads the data to the planning grid in edit mode.  The data is at the same level of aggregation that the monthly stat job is at: Material, Plant, Customer.
• The planner then clicks the composite forecasting button (Ctrl+F10), and selects the same master forecast profile that is used in the monthly stat batch job.
• The system produces the 'Forecast in Interactive Planning' screen and shows the stat forecast, history, composite ex-post forecast, and the forecasts yielded by the 4 univariate strategies grouped in the composite forecast model.  The stat forecast row is populated with the forecast strategy that had the lowest RMSE out of the 4 univariates, lets call this forecast A for illustration purposes... however... the planner doesn't like this forecast and wants to select 1 of the other 3 options available.
• The planner highlights the row that contains the forecast that they want and clicks the 'Adopt single result' (F9) button... this changes the stat forecast row above, adopting the values from the selected forecast strategy.  Lets call this forecast B.
• The planner then saves their actions by clicking the save button at the top of the screen, and green arrows back to the planning book (/sapapo/sdp94), clicking save if prompted.
• <ISSUE> when the planner ends up back in the planning book, the selected forecast (forecast B) is not present, the system selected forecast based on error reduction is (forecast A). <ISSUE>

Has anyone run into this issue where when using composite forecasting interactively, the adopt single result function doesn't save?

Hopefully the example is clear enough, any help, advice or experiences you could offer would be greatly appreciated!!

For reference, the standard userid parameters we've implemented are attached, thanks again for your help.

[rupesh_brahmankar3 · Accepted solution]

Hello,

Could you please check OSS note     1845385 - Composite Forecast not adopting single result.

for lifecycle planning defined for the respective products and the aggregated lifecycle planning is set.

Best Regards,

R.Brahmankar

[christopher_smith1 · Accepted solution]

Thank you sir, LCP is definitely the culprit causing this issue... I'm embarrassed I missed that note, thank you for the info.  In the process of implementing the correction and will let you know

thanks again!