Published on
Why Spend Months Building a Forecast Model if the Next Crisis Is Unpredictable?
Why spend months building a new forecasting model if the next crisis cannot be predicted anyway?
Malte Horstmannshoff of Sartorius answered that question clearly at the C4B live event in Hamburg.
No bottom-up process and no machine learning model would have seen Covid coming. That is fine, because one thing matters above all: being able to set up a new scenario simulation quickly, as soon as the old forecast no longer holds.
Sartorius ran the new model in parallel with the existing process for months. Deliberately, to build trust.
In the end the combination of AI forecasting and simulation clearly beat the old process on accuracy. With traceable assumptions instead of a black box.
Because the process now takes less effort, it also runs more often.
What matters is how quickly you can recalculate.