Fast-and-frugal trees are classification trees with sequentially ordered cues that aid in decision making. Fast-and-frugal trees (FFTs) are very simple illustrations of heuristic decision making. Each tree is comprised of sequentially ordered cues – or questions. In turn, each cue has two branches according to how the question can be answered:
- If the answer to the question is yes, then the branch leads to the next question in the sequence.
- If the answer to the question is no, then the branch leads to an exit point in the sequence.
At the final cue in the sequence, both branches lead to an exit point to ensure that a decision is made either way.
Why is the fast-and-frugal tree useful in business?
Fast-and-frugal trees are particularly useful when decisions need to be made quickly. They are well suited to binary classification problems – or problems with elements occupying two possible outcomes.
FFTs have been trialed in emergency room scenarios to help physicians triage patients. During a peer-reviewed study, a classification tree of just three cues enabled doctors to diagnose and then direct patients to either a regular nursing bed or the coronary care unit.
Cues were based on historical acute heart disease data, allowing high-risk patients to be identified quickly and accurately. In fact, the process was so accurate that it was a better predictor of heart disease than the clinical judgment of the physicians themselves.
Constructing fast-and-frugal trees
There are several ways to construct fast-and-frugal trees.
In the emergency room example, physicians had historical data on factors that lead to acute heart disease. Chest pain was one such predisposition, leading to the creation of a cue entitled “Chest pain chief symptom?” with a yes or no answer.
Although FFTs were designed to be simple, they have nonetheless been adapted by using more complex methods. Primarily, this is seen in FFTs that are constructed using a zig-zag algorithm.
Here, the tree is created using positive and negative cue validity. This validity is defined as the proportion of cases with a positive/negative outcome in all cases with a positive/negative cue value. Typically, the first – or “root” – cue of a zig-zag analysis tree is the cue with the greatest positive (or negative) validity. This ensures that the most significant positive and negative decisions are made first.
Typically, cue validities are determined by using counts and ratios. In more advanced scenarios, they must be estimated using elements of probability theory such as conditional independence. Zig-zag decision trees enhance the already strong fundamentals of FFTs. Given that the tree can be completed with pen and paper, the accuracy of a zig-zag tree is as high as using a logistic regression model.
Fast-and-frugal tree applications
As noted, FFTs are useful in any situation requiring fast and accurate decisions or risk assessment.
Beyond medical applications, these trees have been used in the military to identify enemy threats and also in courtrooms to decide whether to bail or jail a defendant.
In recent times, fast-and-frugal trees have also shone a light on how customer management decisions are made. Retail banking sales managers who embodied fast, frugal, and adaptive decision making were able to better anticipate client needs.
- Fast-and-frugal trees are heuristic models that are useful for tasks where binary decisions or classifications need to be made.
- Fast-and-frugal trees are made of sequentially ordered cues, otherwise known as questions. Each cue has a binary answer, with one answer leading to a subsequent question and the other leading to an exit point.
- Fast-and-frugal trees are simple and effective decision-making tools. However, the decision-making process is enhanced by using historical data and aspects of probability theory.
Connected Thinking Frameworks
Main Free Guides: