Automated methods of mass appraisal
The second seminar on automated methods of mass
appraisal and market analysis: And the discussion moves
on…
This event took place on Thursday November 1, 2007 at OTB Research Institute for the Built Environment, Delft University of Technology, Delft, The Netherlands.
Goals of the seminar:
1. In an educational event examine the market base of property value and processes shaping it.
2. Presentation of our forthcoming book Advances in mass appraisal methods – an
international perspective.
In the first seminar, held in the same place on October 30, 2006, the aim was to understand the problem of advancing mass appraisal methods/expertise based on conceptual ideas and empirical evidence. In this project we compared and assessed several approaches, some of which may be considered advanced (e.g. hierarchical modelling) and others emerging (e.g neuro-fuzzy and genetic-fuzzy modelling). The primary performance criterion was the accuracy of value prediction and other criteria, the benchmark being the linear, parametric MRA. Interestingly, the relative accuracy performances of the methods varied across the datasets. In fact, using datasets that are more heterogeneous in term of the composition of the property, location and environmental characteristics (several house types, varying features of the nature and population etc), the more advanced methods tended to outperform MRA, but using a more homogeneous set the advantage in favour of the advanced method was smaller or non-existing. In a few cases the result was the opposite: MRA outperforming the more advanced method.
This indication of context dependent market modelling performance is potentially a powerful finding that moves the discussion to a higher level. At the same time this finding brings added complexities to the analysis when a market needs to be classified based on its particular features – tangible as well as intangible ones. Furthermore, the market structures are being shaped by institutional and behavioural processes that involve considerations of a
differentiated and partly qualitative nature. Such a methodological framework not only would provide the basis for a more scientific market analysis, but it would also then comprise natural tools for multidimensional sustainability assessment. Sustainability is defined as a long-term criteria for development that has (at least) social, cultural, ecological, environmental and – indeed – economic dimensions. So there is also an economic sustainability – or sustainable market/price! Arguably sustainability is required for mass appraisal and other property valuation too.
adjusted away from ‘modelling tools’ – the last year’s theme – towards the ‘market itself’. The topic of advancing mass appraisal, value modelling and market analysis will be enhanced towards understanding and explaining market processes and institutions with particular interest in the location and spatial dimensions. At the same time the focus will remain in empirical, quantitative applications that look for estimation or classification of a group of objects (dwellings, locations, areas, property types, land uses etc) for special purposes. The common denominator between the contributions is the spatial-market oriented aspect of the property industry.
The seminar has two parts. The first part comprises seven presentations on the analysis of various market contexts and externality effects (each 20 min + 10 min discussion). The second part of the seminar is to promote a forthcoming book published by Blackwell (RICS Real Estate Research Series) that includes the proceedings of the last year’s event.