Evaluating Design Options against Requirements: How Far Can Statistics Help?

Abstract

Trading-off candidate designs against requirements is a critical activity for many projects. This is especially so where the goals of many stakeholders conflict, and therefore cannot all be satisfied. Traditionally, weighting has been used to try to combine scores on different criteria, so as to identify a winning design. However, this has a weak mathematical basis: criteria should be independent dimensions, and may be measured in different units. The statistical technique of Principal Components Analysis offers a robust approach: given clear data, it gives clear guidance, of the form: "if you prefer these criteria, you should favour these candidates". Otherwise, it indicates that no guidance can be given. Either way, this rightly places responsibility for decision-making on human shoulders. The outcome is an improved trade-off process for projects.