Friday, April 3, 2009

Working Smarter With Decision Trees

Summary

This article provides an introduction to using and creating a decision tree to visually demonstrate the process of a particular decision. This tool allows the user to create a model of the relationship of a decision, strategy, and possible outcomes. Decision trees are not only a useful asset to analysts, but are also good tools for presenters and audiences.

At the most basic level, decision trees begin with a goal and the strategy that is most likely to achieve it. The analyst should begin by formulating the main problem, then logically structuring alternative strategies and their projected results. Once the problem and the alternatives are determined, it is now time to construct a decision tree diagram.

There are four main components used in creating a decision tree.
  1. Trigger Event or Decision Statement (Nodes)
  2. Decision Paths
  3. Chance Points or Nodes
  4. Leaf or Terminating Nodes
**For each node, the article recommends using a different shape or color to distinguish between decisions, chance points, and leaf nodes.

Outcomes may flow from either left-to-right, or from top-to-bottom, so when creating a decision tree, begin from the top or the left-hand side. The first step is to identify the trigger event, or decision statement, and place it on the top or left side of the page. The article uses a Product Line Decision as a sample for this exercise and "Extend or Consolidate" as the decision statement. Since there are two separate options for this decision, both pathways branch out from the statement. The decision paths then run into subsequent decision points, resulting in the development of more options. The decision points continue to fan out all probable pathways until they reach the leaf or terminating node. Here is the Product Line Decision Tree Example:




No comments:

Post a Comment