Article ID: | iaor20041615 |
Country: | United Kingdom |
Volume: | 54 |
Issue: | 10 |
Start Page Number: | 1050 |
End Page Number: | 1058 |
Publication Date: | Oct 2003 |
Journal: | Journal of the Operational Research Society |
Authors: | Ward S., Chapman C. |
Keywords: | risk, artificial intelligence: decision support, project management |
A ‘constructively simple’ approach to estimating uses a decision support modelling paradigm based on project risk management and operational research concepts. It employs probability models selected from a set of alternative stochastic models of uncertainty with a view to maximising the insight provided, given an appropriate level of complexity. It addresses issues that include the joint use of subjective and objective probabilities, subjectivity of model data and structure, bias, data acquisition costs, the importance of getting an estimate right, optimising the estimating processes involved as a whole in approximate but robust terms, and differences in interpretation of what this means to estimators and users of estimates. Specific applications are necessarily context-specific to some extent, but the underlying ideas are of general applicability. This paper uses a simple example involving estimating the uncertain duration of a project activity to illustrate what is involved.