Parametric estimating

Parametric estimating generalises the factored approach: instead of one ratio to equipment cost, it fits cost-estimating relationships to whatever parameters the historical data shows to be predictive — capacity, area, throughput, weight, power. The relationships range from a simple unit cost multiplied by a driver to multi-variable models with capacity exponents, and they are applied when the parameters are known but the design is not.

The method is only as good as three things: the size and relevance of the underlying data set, the normalisation applied to it (time, location, scope boundaries), and the honesty about whether the new project sits inside the range the data covers. Extrapolating a relationship beyond the largest project it was fitted on is the classic failure — cost rarely scales linearly, and the exponent that was true from ten to fifty units of capacity has no obligation to hold at five hundred.

Used within its range, parametrics are fast, cheap and defensible, which is why screening studies, benchmarking checks and independent reviews all lean on them.