RISKOptimizer Clarifies Uncertain Choices
By Peter Coffee
March 22, 1999
RISKOptimizer, released last month, will feel familiar to anyone experienced with Palisade's long-established @RISK (the product that practically invented the genre of uncertainty modeling add-ins when Lotus 1-2-3 was on almost every PC). To develop an uncertainty-based model, we entered formulas using Palisade's probabilistic functions such as RiskNormal (for a Gaussian normal bell curve) with additional parameters such as mean and standard deviation.
Stands firmly on uncertain ground
To perform uncertainty-based optimization with RISKOptimizer, we had to define a fitness function: a formula in our spreadsheet that RISKOptimizer could use as a measure of success, either by maximizing (as with an income) or minimizing (as with a cost).
Importantly, however, RISKOptimizer also gave us the option of looking at fitness-function statistics from several trials so that we could, for example, seek a combination of values that produced the most consistent possible result (by minimizing the fitness function's variance or standard deviation).
It was easy to include common-sense considerations such as values that could only be integers. With a formula such as =ROUND(RiskTriang(3,7,15),0), we could concisely describe a quantity that is never less than 3, never more than 15, probably close to 7 and unable to have a fractional value. RISKOptimizer also capitalized on other Excel capabilities, for example by letting us execute named macros at selected times (such as at the end of an optimization process, enabling us to automate report generation).
Excel's Visual Basic for Applications can be used to control RISKOptimizer's own large vocabulary of macro commands. A custom application developer should be able to hide RISKOptimizer behind a custom-built front end, using high-level user input to construct and execute a corresponding model.
An interactive tutorial walked us through a representative RISKOptimizer problem, presenting a realistic decision-making situation involving the overbooking of an airline flight. This clarified options and procedures that could otherwise be opaque to anyone but an operations research specialist.
Executive Summary: RISKOptimizer 1.0
RISKOptimizer 1.0, Palisade's add-in for Microsoft's Excel, works with uncertain quantities and offers an exceptional variety of optimization criteria while sharing the easily documented modeling format familiar to users of Palisade's @RISK.
+ Formula-based notation makes
all assumptions apparent; interactive tutorial clarifies many topics
that will be new to most users.
- Interactive tutorial falls short in explaining underlying genetic optimization algorithm.
Related Article in the March 1999 PCWeek:
» Why Plain Old Spreadsheets Are Not Enough
For more information, contact:Randy Heffernan