ABSTRACT
Predictive modelling is a set of techniques, used since the 1970s, to predict the location of archaeological sites in uninvestigated areas as an aid to spatial planning, for example, in Cultural Resource Management. Predictive modelling is also used to develop and test scientific models of human locational behaviour, as it is based on either statistical extrapolation from known archaeological data, or on explanatory models of site location preference. In practice, a number of methods can be used in predictive modelling, and the resulting maps of predicted site locations or density can vary in accuracy. The main difficulties in producing accurate and precise predictive models are coupled to the resolution and representativeness of the archaeological and non-archaeological datasets used, the theoretical frameworks underlying the models, and the nature, or lack, of model testing. Nonetheless, predictive models are often found useful to provide basic protection to areas of high sensitivity, and can save costs for development projects or archaeological investigations.
