Decision Analysis Theories

This apparent inattention to an important discipline disguises a bewildering clutch of widely unused and mainly academically devised approaches to decision making that are available and categorized under the broad-based classification of Decision Analysis. They include techniques such as Analytic Hierarchy Process (AHP), Bayesian Updating, Outranking, Subjective Judgment Theory, Utility Assessment, Matrix, Cost–Benefit Analysis, and the Decision Tree—the last three being the most commonly known.

Matrix, for example, utilizes a subjective weight assignment for alternative criteria, its main drawback being that it cannot account for interdependence between so-called best alternatives. Cost–Benefit Analysis, which provides a quantitative format for reckoning the range of benefits and costs surrounding a prospective decision, aggregates the effects over time using an approach called discounting, and arriving at a “present value” or “payback period.” In its simplest form, it is carried out using only monetary costs and benefits, but a more sophisticated approach tries to put a financial value on intangible costs and benefits, which makes the calculation highly subjective. Other criticisms include the imprecise techniques used to measure diverse benefits and costs and the fact that, to some, environmental concerns fall properly under the realm of ethics rather than economics.

The Decision Tree is an abstract methodology in which alternative decisions and their implications can be evaluated via a genealogy-type visual aid. Its main criticisms include overfitting—when the tree matches random variations of the target values in the training data that are not replicated in other samples—and instability—when the tree fits the data well, predicts well, and conveys a good story, but then, if some of the original data is replaced with a fresh sample and a new tree is created, a completely different root-and-branch picture may emerge using completely different inputs in the splitting rules and, consequently, conveying a completely different story. However, as loyal as followers of any of these techniques might be, the historical shortcomings of industry and commerce confirm that good decision making is a far more practical discipline than these methodologies accommodate.