Suppose that friendship is a 2-way relationship: Either two people are friends with each other, or they are not. (By this definition, X cannot be a friend of Y if Y is not a friend of X. Also, you cannot be a friend of yourself -- no matter how attractive and charming you happen to be.)
Given the above characterization of friendship, I will make a forecast about every group of people:
In any group of 2 or more people, there is at least one pair of people who have the same number of friends in the group.
This is one of those nice situations when I know I will achieve 100% forecast accuracy. Not because I’m conceited about my forecasting abilities or can control the outcome, but because of logic and mathematics.
(Don’t believe this? The first one to send me a counter-example gets a free copy of The Business Forecasting Deal. Or else send me a proof that my forecast will always be true, and I’ll print your name in a future installment of The BFD.)
When perfect (or at least highly accurate) forecasting is the goal, it can sometimes be achieved by controlling the outcome. For example, you could constrain your supply and only sell up to your forecasted sales (at the cost of disappointing those customers who came in too late). This seems kind of a bad thing to do.
Controlling the outcome may not always be a bad thing, however. A university may wish to plan for future enrollment, allowing them to maintain appropriate staffing and facilities, and better manage costs. The university has considerable control over its admissions process, so achieving great forecasts of future enrollment may not be that difficult (or laudable) an achievement.
Coming Next Week: INFORMS in Chicago
Please join me and already over 600 registered attendees at next week’s INFORMS Conference on Business Analytics and Operations Research. The Forecasting track (on Tuesday, April 12) has these topics:
• Quick Service Restaurant Forecasting: Challenges, Methods & Solutions, Rainer Dronzek (McDonalds)
• Econometric Models to Forecast Brokerage Revenue at BoA Merrill Lynch, Russell Labe (Bank of America)
• Large-Scale Statistical Forecasting for Cisco’s High-Tech Demand Patterns, Andrew Fisher and Mert Sanver (Cisco Systems)
• Deciding Which SKUs to Forecast: Simulating Business Outcomes to Reduce Risk / Improve Forecasts, James Hoover (Accenture)
• Prediction Markets: Practical Advice, Secondary Benefits and Surprising Results, Thomas Montgomery (Ford)
Be sure to stop by the SAS booth to say hello and talk about forecasting.
