The decision making tree is one of the better known decision making techniques, probably due to its inherent ease in visually communicating a choice, or set of choices, along with their associated uncertainties and outcomes. Their simple structure enables use in a broad range of applications. They can be drawn by hand to help quickly outline and communicate the critical elements in a decision. Alternatively, a decision tree's simple logical structure enables it to be used to address complex multiple decision scenarios and problems with the aid of computers.
Using a simple decision tree example, we can see the basic elements used when visualizing a choice. The drawing will generally have the following elements:
As seen in the above example the tree will model decision options with their consequences, including uncertain outcomes. In this example, the outcome of investing in ten different stocks has an equal chance of returning $13,000 or $9,000, leading to an expected (or average) return of $11,000. Three decisions are shown, starting with the choice to invest $10,000. A tree will also show all possible outcomes as exemplified by the investment returns of $13,000, $9,000, and $10,300.
To use the tree, a common measure of utility is used that enables determination of the path through the tree. As in this example, often utility is measured in dollars, making it easier to evaluate the impact of costs associated with some of the choices (e.g. the stock purchase cost).
The decision making tree is usually built starting with the initial decision option, and moving through choices and chance events until all outcomes are reached. Once the tree is developed, you work backward from the outcomes to determine the values used to find the best path or set of choices to move through the tree. In this case the expected return for the stock purchase of $10,900 makes it the best choice for the $10,000 investment (although with greater risk).
Decision trees have a natural "if ... then ... else ..." construction that makes it fit easily into a programmatic structure. They also are well suited to categorization problems where attributes or features are systematically checked to determine a final category. For example, a decision tree could be used effectively to determine the species of an animal.
As a result, the decision making tree is one of the more popular classification algorithms being used in Data Mining and Machine Learning. Example applications include:
Because of their simplicity, tree diagrams have been used in a broad range of industries and disciplines including civil planning, energy, financial, engineering, healthcare, pharmaceutical, education, law, and business.
One of most systematic tools for decision making theory and application, tree diagrams have been used for many complex multi-stage decision problems. They can help form an unbiased view of risk and opportunity associated with any choice. They are often used in decision analysis to help identify a strategy with the highest likelihood to achieve a goal.
Decision making trees are an effective technique because they provide the following benefits and advantages:
The systematic and logical structure of the decision making tree still requires the decision maker to scrutinize and validate the attributes that are being used to build the diagram. As in all logical structures, if the assumptions are false, conclusions will be misleading. The tree builder (decision maker) must ensure the distinction between causality and correlation, particularly for trees that will be used for predicting future outcomes.
Additional items to consider when choosing to use a decision tree include:
Here is a great example of a decision making tree that extracts and models knowledge gained by experts over a long period of time. The Vroom-Yetton-Jago Decision Model provides an alternative approach for choosing your style for collaborative decision making. This decision making tree packs years of experience into a single page that can be navigated with 8 questions that require only yes/no or high/low answers.
A closely related analysis method is the influence diagram that is also a highly visual decision support tool. Influence diagrams focus on relationships between decision events and can provide a way to compact the information presented in a decision tree.
NPV analysis is often developed and visualized using a decision making tree. The tree diagram helps reveal where key risks are being added to the project being evaluated.
On the internet, a number of resources are available that support decision tree analysis, and often these tools will support influence diagrams, NPV analysis, and other related methods.