IECM 12.0 beta User Manual > Introduction to Uncertainty Analysis >

Philosophy of Uncertainty Analysis

The classical approach to probability theory requires that estimates for probability distributions be based on empirical data. However, in many practical cases, the available data may not be available or relevant to the problem at hand. Thus, statistical manipulation of data may be an insufficient basis for estimating uncertainty. Engineering analysis or judgments about the data may be required.

An alternative approach is the "Bayesian" view. It differs in how probability distributions are interpreted. The probability of an outcome is your "degree of belief" that the outcome will occur, based on all of the relevant information you currently have about the system. Thus, the probability distribution may be based on empirical data and/or other considerations, such as your own technically-informed judgments. The assessment of uncertainties requires thought about all possible outcomes and their likelihood, not just the "most likely" outcome. The advantage to thinking systematically and critically about uncertainties is the likelihood of anticipating otherwise overlooked problems, or identifying potential payoffs that might otherwise be overlooked.