Fast-and-Frugal Trees: A Review

A fast-and-frugal tree (FFT) is a tree that allows for classification at each level (node) of the tree. This means that a tree with K binary cues (attributes) will have K nodes and K+1 exits (end nodes or leaves). This makes... [ view full abstract ]

A fast-and-frugal tree (FFT) is a tree that allows for classification at each level (node) of the tree. This means that a tree with K binary cues (attributes) will have K nodes and K+1 exits (end nodes or leaves). This makes fast-and-frugal trees much simpler than the complete trees with 2^K exits. Indeed, due to the exponential increase of the exits with number of cues, complete trees quickly become computationally intractable for large number of cues, and this makes FFTs operationally attractive and more robust. In an FFT, cues are usually ordered according to one-reason decision-making and one-reason stopping rules (Gigerenzer, 2004, and Gigerenzer & Gaissmaier, 2011). There are a variety of applications where FFTs have shown to perform exceptionally well compared to competing models. This paper, reviews a selection of such FFTs in the fields of Medicine, Psychology, Law, Finance, Banking, and Military Stability Operations and shows how they work in action.