It can be argued that linear regression is not appropriate in LVQ analysis since visual response
to landscape factors is unlikely to be either linear or independent. Other regression modelling
techniques (e.g. Neural Network regression model, Bishop 1996) are available but have methodological
issues of their own. As mentioned, most research in this field simulates LVQ using linear
regression. The Analytic Hierarchy Process (AHP) is also widely used in multi-criteria decision
making (Saaty 1994; Schmoldt et al. 2001). However, AHP or similar methods are not suitable
for visual modelling because of the difficulty people have in identifying the role of individual
factors in their subjective landscape judgements. Consequently, despite its deficiencies, the regression
approach still dominates research and application to visual quality landscape assessments