This is a question that we, as IRIS by Argon & Co, get asked regularly by our clients when looking at their Demand Planning processes: what is the forecast accuracy I should aim for, given my own industrial context? The immediate answer is to consider the perfect forecast, with 100% accuracy, fully aligned with the actual demand.
This target is obviously unrealistic: as the world around us is not fully predictable, the actual sales timeseries will always have a noisy and unpredictable component; hence the forecast accuracy shows a glass ceiling effect, which can be much lower than 100%. So, as a Demand Planner or Data scientist evaluating forecasting models, how do you set your Forecast Accuracy objectives? How do you know when your forecast can be considered
good enough ? Is forecast accuracy even the right metric to evaluate?
Discover the whitepaper below :
Contact : email@example.com