Abstract:
A new approach to mass appraisal in cases when dependent variable contains an unobservable part (due to market immaturity) is proposed. The approach includes several important stages: a) construction of two models with required assessment quality and defined on the respective non-overlapping sets (classes); b) classification of assessed objects; and c) selection of appropriate model to assess each object. The approach allows to better use the market information and improves the quality of mass appraisal.
Keywords:regression model, $d$-vicinity, classification, accuracy, reliability, mass appraisal.
Fulltext:
PDF file (861 kB)