Getting to the root of decision-tree analysis: a clear explanation

An advertising supplement in the ACC Docket, Vol. 27, Jan./Feb. 2009 after page 48, very clearly describes decision-tree analysis. Brian Daley BDaley@ogilvyrenault.com of Ogilvy Renault writes:

“At its most basic, a decision tree is a diagram that starts with two branches that represent a choice, for example to litigate or settle a lawsuit. Each branch expands into a series of sub-branches tht represent all of the possible occurrences associated with each choice. The sub-branches end when a final outcome is reached, for example, the defendant is found liable, the defendant is found not liable, etc.

By assigning probabilities to all of the possible occurrences and dollar values to all of the potential outcomes, you can determine and calculate the probabilistic value of each choice. Commercial software products automate this process and provide lots of bells and whistles.”

That tells it all, and clearly, about decision-tree software (See my post of Jan. 17, 2006: decision trees; Oct. 24, 2005: decision trees; June 18, 2007: belief nets and decision trees; and April 22, 2008: quantitative analysis of litigation.).