Multi-Criteria Site Selection

Weights express how much each criterion matters and must sum to one. A common, rigorous approach is the Analytic Hierarchy Process, in which the decision-maker compares criteria two at a time; the principal eigenvector of the resulting comparison matrix yields the weights, and a consistency ratio flags incoherent judgements. Simpler rank-sum, rating, or entropy methods are also used, and weights can be elicited from multiple stakeholders to capture differing priorities. Because the weights are subjective and influential, they should always be tested with a sensitivity analysis.

Criteria are factors that vary continuously and trade off against one another — closer is better, cheaper is better — and are combined through weighted overlay. Constraints are absolute, non-negotiable rules that simply make a location eligible or ineligible, such as 'not on protected land' or 'not within a flood buffer'. Constraints are applied as binary masks that zero out unacceptable locations, so no amount of good performance on the criteria can rescue a site that violates a hard rule.

Weighted linear combination adds up weighted criterion scores, so a high score on one criterion can compensate for a low score on another — a site can rank well overall despite being poor on a single factor. That is fine when criteria genuinely trade off, but dangerous when a low score should be disqualifying (for example, a site cheap and central but on a fault line). In those cases the weak criterion should be handled as a constraint or with a non-compensatory aggregation rule such as ordered weighted averaging or an outranking method rather than simple weighted overlay.