Structuring Assumptions in Decision Making

Abstract

This note describes how one would structure assumptions using influence diagrams and decision trees. It discusses modeling under certainty and under uncertainty. Contrast and comparison between influence diagram and decision trees are provided.

Excerpt

UVA-QA-0499

Rev. Aug. 20, 2009

STRUCTURING ASSUMPTIONS IN DECISION MAKING

Decision making involves knowing your alternatives and having a way to measure their performance. In only the simplest situations can we directly determine the performance level of all alternatives. Generally, we make assumptions about the linkage between a specific alternative and its level of performance.

This note is designed to help students get started in structuring assumptions about how alternatives connect to performance. This involves thinking about what quantities are useful in establishing this linkage, and envisioning what relationships should exist among these quantities.

The note will describe the use of an influence diagram, which shows what quantities are involved, how they relate to each other, how they are affected by alternatives, and, in turn, how they affect performance. The influence diagram is helpful for framing assumptions and can ensure that a messy situation has been brought under control. It can provide the structure for building a spreadsheet model and for assessing the quantities that go into it. Presenting the influence diagram can be more effective than displaying the mathematics of a decision model or the background of a spreadsheet. New software tools use this diagram as a graphical interface for creating, manipulating, and presenting the assumptions underlying a decision model.

. . .

Keywords: influence diagram, decision trees, decision analysis, structuring, modeling