Demand Foresight

BLOG

Forecast Modeling Matters

Your forecast is the key to getting your expenses and customer satisfaction in line.

Forecasting IS the first step in managing your expenses, a good forecast can mean the difference between too much inventory and upset customers. Not to mention keeping supplier shipment costs as low as possible, nothing like air freighting from China!

There are a lot of statistical forecast engines on the market, ranging from a few hundred dollars to thousands. The biggest issue we have seen with pure statistical models is they are math, they don’t account for the real-world business considerations.

For example have you seen a forecast trend down

This is a classic example of a pure statistical forecast not understanding the real-world as applied to business.

After years of working in the forecasting planning area we have come up with our own proprietary modeling engine. Here with the same data is how our Forecast Engine models with no user intervention, this is through training to get the lowest forecast error (MAD, MAPE, BIAS etc..)

Take this example, you have an unexpected spike. Here is how most statistical models would handle this.

This is how our out of the box Forecast Modeling Engine would address this issue.

We have taken our modeling a step farther where our models compete for the lowest error. We don’t have one or two models to use we have hundreds of auto competing models to find the lowest error all using our proprietary business intelligence.

But to go beyond this we allow you to “Restate” history when you know there was an anomaly as well, like a new warehouse stocking that artificially drove up shipments/bookings.