[ Go to October 1997 Table of Contents ]|
-- by Martin Heller
Sometimes you need more sophisticated data-analysis tools than the features your spreadsheet can provide. To meet these needs, Decisioneering offers Analytica, a visual modeling tool that can handle complex tasks while requiring less expertise than most other computer-modeling methods.
You build an Analytica model by drawing influence diagrams and then defining the objects in the diagram. Spreadsheets allow you to conduct a sensitivity analysis by systematically changing variables in a table and tabulating or graphing the results. You can introduce tables and statistical distributions for any variable. Analytica automatically tabulates and plots the results.
The simplicity of drawing diagrams belies the complexity of quantitative analysis. If a factor varies statistically, it's important to understand its underlying distribution to obtain realistic probabilities. Analytica includes a menu of distributions and probability graphs, but it doesn't teach you which ones are appropriate for any given scenario. In other words, without a little background in probability and statistics, you are not likely to be comfortable or effective using Analytica.
The beta version we tested has only recently been ported from the Macintosh. It lacks online help and supplies only Macintosh printed documentation. Nevertheless, we found Analytica remarkably easy to use. It also performed well; on a 200MHz Pentium Pro, it was fast enough for complicated models.
Analytica has little competition, since it is much easier than modeling by programming and more flexible than modeling with a spreadsheet-even one enhanced with Decisioneering's Crystal Ball forecasting and risk-analysis tools. If you need to perform modeling and quantitative analysis, and you know how, Analytica is a good choice. But if you don't, no tool will substitute for expertise.
Windows Magazine, October 1997, page 160.