Multicriteria Analysis

Multicriteria analysis is a set of analytical models of individual or group decision making when multiple considerations, possibly conflicting with one another, are at play. Systematic procedures for analyzing complex decision situations have been researched and developed since the 1960s, primarily in the fields of management science, military science, and operations research. Geography, particularly analytical economic geography, has become a major testing ground for the validation of multicriteria decision techniques owing to the fact that most data used by individuals, as well as by public and corporate managers, have a geographic component.

In some instances, models of multicriteria decisions serve to comprehend the reality of the process and factors leading to making decisions. Utility-based models of residential choice, in which trade-offs exist among multiple characteristics of housing units as well as the relative location of each choice alternative with respect to employment, retail, and service districts, are classical cases of such a descriptive framework. On the other hand, multicriteria analysis can also set a standard to which decision making ought to adhere. A state agency that is looking for a route to ship a truckload of hazardous materials to a processing center would rely on a normative approach to minimize the risk of population exposure while maintaining travel time and travel cost within acceptable bounds. Both classes of decision problems abound in spatial planning, business operations, and public policy and management.

A multicriteria decision problem typically involves multiple components, including a goal or set of goals, a decision maker or group of decision makers, choice or decision alternatives (e.g., actions, locations, policies, objects), and criteria against which alternatives are evaluated. A criterion is a rule on which to test the desirability of alternatives. When criteria are the attributes of geographic entities or decisions (e.g., distance, population density), the phrase multiattribute decision making (MADM) has been used. On the other hand, when criteria are identified to the decision makers' objectives with respect to the desired state of the outcome of the decision process, an analysis of multiobjective decision making (MODM), whereby objectives are functionally related to attributes of the alternatives, is conducted.

Even when multicriteria analysis is intended to replicate the outcome of human decision making, its methods are merely a simplification of the real processes of decision making used by humans. The decision process typically is decomposed in a small number of manageable tasks. A common framework for MADM would involve the following steps: definition of the decision problem (What is to be accomplished?), compilation of decision alternatives (What are the options?), identification of evaluation criteria and of their relative importance to the decision maker (What is the decision to be based on?), evaluation of each alternative on each criterion (How does each option fare from different perspectives?), integration of partial evaluations into an overall performance score of each alternative, and (finally) option selection. Multicriteria analysis now is routinely interfaced with geographic information systems (GIS) to fully incorporate the inherent spatial dependencies among alternatives, their attributes, and decision objectives.