Friday's quote of the day:
"The uniform distribution is the mathematical representation of ignorance."
Jack Soll, describing how a set of assumtions can be used to build models used to forecast future events. Our class was analysing the Eurocomp case, where a manager had to decide whether or not to enter the market for computer hard drive manufacturing. In doing so, he had to evaluate three things:
-
Predicted demand for hard drives
-
His internal cost of production
-
The market price of hard drives two years into the future
For #2 and #3, our trusty manager predicted a normal distribution, and gave corresponding means and standard deviations. For #1, he threw up his hands and said it was equally likely that demand would fall in a given range (thus the uniform distribution).
This was the class in which we learned about covariance and the coefficient of correlation. Coincidentally, in the case, the three estimates were defined as independent, so when our professor asked if these estimates were reliable, I pointed out that items #1 and #3 were negatively correlated, and that should be represented in the model. It helped me a lot that making models is what I did for a living for the last few years...
9:43:07 PM
|