SAP IBP- Improving accuracy reporting with WMAPE

SAP IBP defines additional metrics to evaluate the forecast quality of the statistical forecasts compared to what we had in SAP APO. The biggest improvement in IBP Demand is addressing the MAPE     

A flawed metric that calculated infinite MAPEs when you have zero demand for one month.

IBP introduces the WMAPE

Weighted Mean Absolute Percent Error weighs the individual observations by the actual volume. Thus, zeros do not create the funny behavior of leading to infinity.

WMAPE calculates the absolute deviation over the ex-post period and divides by the sum of the actual volume. So, individual zeros do not matter.

Weighted Mean Absolute Percentage Error

What a relief!!

MAPE was useless in SAP APO particularly when your time series had intermittent demand. And is still useless in IBP when you have intermittent demand pattern.

So one should really choose WMAPE as a measure of accuracy and discard the MAPE. Ironically, there is no equivalent measure to resolve issues in the MPE - Mean Percent Error.

What is SMAPE?

                 

                 Symmetric Mean Squared Error

For a detailed set of metrics and how to calculate them, please contact us