Statistical Forecast Qty Accuracy calculation

[former_member594111]

Hi Experts,

Could you please tell how Statistical Forecast Qty Accuracy(%) is calculated?

In manage forecast error calculation app

Input key figures are-Actuals Qty and Statistical fcst Qty(1M lag)

Output key figures are-Statistical fcst Qty Bias(%) ,Statistical fcst Qty Error(%) and Statistical fcst Qty Accuracy(%)

In template i ran forecast error operator but nothing changed in template Could you please tell why this happening?

and What is significance of Statistical fcst Qty(1M lag), how its calculated?

Regards,

Abhijeet

[lingaiahvanam · Accepted solution]

Hi Abhijeet,

You need to follow the below steps.

1. You need to set up the forecast key figures based on your requirement

2. Use the COPY operator to populate the key figures STATFORECASTLAG1 key figures (prerequisite is run the statistical forecast process or upload data into lag key figure )

3. Run the forecast error job

For more information check this Forecast Error

Best Regards,

Lingaiah

[former_member591269]

Hi Abhijit,

I Couldn't see any fault in the Parameters selection in the Copy operator. Please check with Dev team.

Thanks & Regards,

Pradeep K C

[former_member594111]

Hi Pradeep,

Please check below screenshot of copy operator

[Irmi_Kuntze]

Did you read the info which was posted with the link?

You will find there that the FC error calculating the way it is set up e.g. in the UNIPA panning areas calculate comparing the actuals versus the forecast. For that, you need to have forecast and actuals in the same periods, e.g. you made your forecast three weeks ago for last week, and today you are able to compare the forecast values with the than available actuals. If you have your very first forecast calculation today, you only have future forecast values (except when working with offsets), that can not be compared with actuals that have already happened in the past.

And that is why you need the lag figures: To capture the forecast with lag 1 or lag 3 or whatever you want to compare later with the real actuals. Business target usually is to monitor the accuracy of the actual forecast, and validate whether the different sources of input add business value to the forecast

If you want to understand the forecast error directly with the stat. FC execution, you could use the functionality to run ex-post-forecast and let the error be calculated on the last tab of the FC modelling. In that case, the ex-post-forecast (which is in the past) can be compared to the actuals. That is usually done in order to test the algorithms that are used for forecasting, and make sure that the best ones are used.

[former_member594111]

Thank You Lingaiah for your response.

I used copy operator to copy value from statistical fsct qty to STATFORECASTLAG1 but values are not copied.

[former_member591269]

Hi Abhijit,

Have you maintained required Parameter's (Example Attributes, duration, PeriodID, Period_Offset, Source and target KF etc) in the Copy Operator?

Can you share your copy operator screenshot in the group.

Thanks & Regards,

Pradeep K C