How to Use a Decision Tree

A decision tree has many applications. Although it is a common tool in diagramming program flow and calculating conditional probability solutions, decision trees can be utilized anytime a series of decisions leads to an outcome or potential outcomes. A decision tree is a graphical representation of each decision, along with each possible response to the respective decision. Along the branches of the tree can be additional data, often value-based data---such as cost or other factor associated with a particular course of action.
Use a decision tree to analyze most aspects of a business's operation or to understand what is going on with a given set of data. The decision tree gives you a graphical representation to study.

Instructions

1. Organize your data. If your decisions are linked to any values---such as costs, time and resources, organize your data in a table. If your decisions funnel a starting point to more than one outcome point, cluster the data. For this, write the various expected outcomes, giving each space. Then around each outcome, list related information: particular decisions directly leading to that outcome, probabilities and other factors. How you organize your data will depend largely on the type of data you are dealing with. Your own personal preferences in logical structuring will also come into play.

2. Determine and isolate the primary decision, the root of your tree. Think of this as the main question you are trying to answer or your base starting point. Draw a square to represent this decision. Note: Squares (diamonds in flowcharts) represent a decision (or question) that has known results/answers, while a circle is commonly used for uncertain circumstances that produce uncertain outcomes.
   Plan your tree to be drawn either from top to bottom or from left to right.

3. Add branches to your decision square and label each with the various options (or results). As well, you can add in value data. Common values are estimated percent probabilities that an outcome will result. But the value can be anything---the day of the week, a weather condition, a monetary amount and a length of time.

4. At the end of each branch, you will either reach an outcome and the branch ends or you will have another decision to make. Use either a circle or square (as is appropriate) for each decision and draw branches for each outcome. Weigh the branches (with values) as appropriate. Repeat this until you have thoroughly covered all decisions and arrived at all possible outcomes.

5. Study this visual representation of your data. Follow the decisions, tracking those with the greatest probability to arrive at the most likely outcome. Note, this does not mean other outcomes cannot apply. It is simply a look at what will generally be expected. If another outcome is more desirable than the most likely one, work backward through the tree. Start at the desired result and go back to study those decisions that directly lead to the more desirable outcome. Ask yourself if there is something in those key decisions that can be changed to increase the probability of the desired branch(es).
   Likewise, do a study of any negative (undesirable) outcomes and back-trace to see where decisions leading to these outcomes can be steered toward better outcomes. Ask yourself what can be done to reduce the odds of those branches being followed. Use the decision tree to better understand the chain of events so that you can better guide those events.

References

• Mind Tools on Decision Trees
• Time-Management Guide on Decision Trees