Friday, November 2, 2007

Forecast Tracking

Okay, so you've made a forecast. Great. Does that mean that we can file it away and forget about it? NO. That would leave you reaching for the panic button when you realize that you missed all of your projected values. Forecasts are most useful as tools when they are constantly used, updated, and modified with the most recent information. It is very important that you monitor and control these forecasts to see how accurate you forecast is and how it can be improved.

To monitor how well the forecast is predicting actual values, we need to calculate a tracking signal. The tracking signal is calculated with this equation:

Tracking signal = RSFE / MAD

where RSFE is the Running Sum of Forecast Errors
and where MAD is the Mean Absolute Deviation.

In other words, the signal tells us whether we are tracking positively or negatively to our forecast. Positive tracking means that we are above our forecast. Negative means that we are below. If we are using a computer adaptive smoothing to do the modeling for us, the computer can automatically adjust our forecasts for us to reduce spikes in the forecasting below or above our model.

If the values are consistently off, that means that we have "bias" in one direction or another. However, don't hit that panic button yet. We we may want to consider a different forecasting model. Following the theory of Focus Forecasting, we should choose whatever forecasting model is best. It is okay to change models frequently, as long as it is cost effective and the new model is more accurate than the last.

Okay, that about wraps up my series covering forecasting. Thanks for reading. I will continue in my next entry with a summary of what I hope you have learned.
------Sincerely, Trevor Stasik.

To return to initial post about forecasting, click HERE.
To go to the final post in the forecasting series, click HERE.

View Trevor Stasik's profile on LinkedIn

Post a Comment