SAP CAR Unified Demand Forecast KPIs and Accuracy

[graeme_williams]

I currently having a look at what we can do to measure forecast accuracy and forecast KPIs from the UDF, so far I haven't found much helpful material other than using holdout forecasts / diagnostic forecasts and looking at the lowest level of material | plant combinations. I'm looking to do reporting on accuracy using MAPE / MAD / WMAPE / bias type measures at all levels of product hierarchies and location hierarchies. This will be used for both management reporting and as guidence to demand planners on which forecasts to focus on for improving accuracy.

Volume is probably near 6m material | plant combinations weekly

We could probably do this by writing to a table of forecast vs actuals over time and calcualte the metrics from there in a SAC report or similar.

Are there any pointers or suggestions in this area?

[tonyhayes]

Graeme,

Did you manage to progress this issue.

Thanks

Tony

[graeme_williams]

tonyhayes - we didn't find anything pre-rolled so we have developed our own in SAC now that is similar to our legacy forecast metrics (calculates MAPE, WMAPE, MAD, and bias at any level of product and location hierarchy), we are quite interested to benchmark the UDF forecast metrics against our legacy tool! We just have to get a couple of cycles in production under our belts before we can get soem reasonable data out.