Fuzzy decision support systems are proven to be very effective in imprecise and incomplete environment. However, the amount of uncertainty associated with the output of these fuzzy systems is never quantified and utilized in decision making process. A new percentage score based tool is introduced in this work to capture this valuable information. By utilizing this tool, it is possible to interpret the confidence of the mechanism on its final recommendation. This allows for the enhancement of information quality and preservation of the uncertainty throughout the decision making chain. Several properties of the proposed output uncertainty score is discussed and proved. Experimentation on real dataset reveals that the output uncertainty depends on the summation of input uncertainty, but does not correlate with prediction accuracy when used in forecasting system.
History
Journal
Engineering Applications of Artificial Intelligence