Call (+1) 650-212-1212


Does Decision Tree Software Soften your Brain?

Sean Salleh 12 Jul 2013 Modeling methods

Rely on anything too much and you’ll end up sacrificing some ability. Kids with calculators are prime examples: ask them to add two and two, and watch those fingers fly around the calculator keypad. It’s a world turned upside-down: machines are supposed to take care of the drudgery and leave us to concentrate on the thought process, not the other way round. Admittedly, decision tree software appeals to a rather different audience. However, is there a danger that over-consumption of such programs can lead, if not to brain flab, then at least to blind spots?

Simple tree diagram example Image source:

Arrows and Arithmetic

According to the creator or vendor of the application, decision tree software can range from the simpler drawing program with an overlay of simple logic and calculation, to a more sophisticated solution capable of dealing with ranges of uncertainty, rather than just a few fixed probabilities. The most basic decision tree applications are easy enough to understand. The program makes sure that the sum of probabilities of different outcomes from a node is always 1 (or 100%), and you can draw out your tree with as many nodes and branches as you like.

Pros and Cons of Using Decision Tree Software

Often you’ll end up with a neater representation on the screen of your decision tree than if you’d drawn it out by hand. And the software will most likely perform calculations of outcomes and payoffs without error. It won’t tell you however if your tree makes sense or if the calculations are the right ones to be made. Neither will any decision tree it draws necessarily show you all the possible relationships between the different factors involved. So to speak, you can roll forwards and you can roll back, but you can’t roll sideways. The result of these limitations on virtual movement is that focus is exclusively on one-dimensional chains of events, and multi-dimensional relationships, usually a closer approximation to real life, are ignored.

Reclaim Your Other Dimensions

If the richness of the relationships between different inputs in a decision making process is to be taken into account, something more than a tree and branches will be needed. The good news (assuming this is important to you) is that you don’t have to give up computer-generated schema. By switching from a decision tree representation to an influence diagram like the ones in Analytica, you retain digital neatness while gaining in clarity and meaningfulness. If your influence diagram becomes more complex, you can organize it into a top level influence diagram showing the major components and their relationships; and underneath a major component, drill down to see the ‘sub’-influence diagram that is driving that component.

Simple influence diagram example Image source:

Don’t Forget Your Bonus

In Analytica, influence diagrams also carry over seamlessly to probabilistic models that take account of uncertainty. You are no longer constrained to apply an invariable probability. You can immediately express that probability as a distribution, and then run simulations with random numbers corresponding to points somewhere along that distribution. A thousand simulations in the purpose built software of Analytica can be done quickly for results that would be impossible to obtain otherwise.

Decision tree software like the handheld calculator has its place: for simple decision questions, for example for accounting procedures, with fixed data or probabilities for situations that need objective treatment or error-free manipulation, for instance. But for multiple criteria and their associated uncertainty, you’re likely to find that influence diagrams and stochastic models rapidly become more relevant and more realistic.

If you’d like to know how Analytica, the modeling software from Lumina, can help you analyze multi-attribute models and more, then try a thirty day free evaluation of Analytica to see what it can do for you.

» Back

Sean Salleh

Sean Salleh is a data scientist with experience in guiding marketing strategy from building marketing mix models, forecasting models, scenario planning models, and algorithms. He is passionate about consumer technologies and resource management. He has master's degrees in Operations Research from University of California Irvine and Mathematics from Northeastern University.

Leave a Comment