Stromy rozhodovania, diagramy vplyvu a analýza rozhodovania
DECISION TREES, INFLUENCE DIAGRAMS AND DECISION ANALYSIS

Summary/Abstract: The article deals with the basic principles of decision analysis. Two tools of decision analysis – decision trees and influence diagrams are compared. Decision analysis provides the structure and methods for systematic analysis of the decision problems. Decision analysis provides along with the framework for thinking about the decision problems also analytical tools, which are able to simplify the decision analysis process. The decision problems of choosing one activity (variant of solution of a real prob-lem) from a finite set of all considerable activities are considered in the article. Decision analysis simplifies the orientation in the decision problems. Decision analysis can be defined as a prescriptive approach, which enables people of average intelligence to think systematically about some complicated, important real problems. Decision analysis provides a certain structural framework and the instructions for sys-tematic thinking in the solution of complicated decision problems. In the application process of the decision analysis often subjective opinions about the uncertainty and about values of some problem parameters are important input data. Accordingly the methods of decision analysis allow to include also subjective opin-ions into the formal analysis. Decision analysis offers also procedures, which facilitate orientation and a certain self-control of the person, his own subjective opinions. Decision analysis is based on knowledge of the statistical decision theory, which deals with decision theory in uncertainty conditions when statistic information is provi-ded, which allows to include some uncertainty aspects into the formal decision problem analysis. Decision analysis can then be defined as the summary of decision methods, which are based on the statistic decision theory. In statistical decision theory the experimental information is combined with other relevant problem aspects with purpose to find the best solution. Several tools for structuralization of decision problems are known. The suitability of application of any of them depends on the character of the particular problem. Some decision problems can be structuralized by payoff table or loss table. This problem and other ones with similar structure can be structuralized by help of the deci-sion tree or the so-called influence diagram. In the article mainly the influence dia-grams are mentioned (less presented in our literature so far) and compared with deci-sion trees on the example of the decision problem known as „prospector problem“. The influence diagram enables a simple graphic projection of the decision problem. It consists of value nodes, decision nodes, chance nodes and arrows, which connect them. In the influence diagram there are mentioned also all options related to the corresponding decision nodes, also all results related to chance nodes together with probabilities of their coming and yields related to value nodes.

• Details
• Contents