Regression Procedures and Best Practice in Real Estate Valuation

An important problem for the mortgage is real estate valuation. A comparative approach to property valuation is the main method of property valuation because its principles are based on market behavior. It requires identification of the impact of various property features on transaction prices. Modelling of that influence is based on principles which are historically conditioned and which frequently do not take into account the contemporary data analysis research methods. Transaction prices datasets generated by broadly understood real estate market are characterized by heterogeneous dispersion and a tendency for local coincidence. This structure induces application of statistical methods based on recreating dependencies in the price set rather than assuming a parametric model ad hoc. In this work we present an assessment of the application of methods based on regression analysis as an example of a parametric method. We indicate advantages and imperfections of using these methods. Specificity of market information is presented which constitutes the basis for property valuation. We propose a canon of best practices for individual valuation and show the advantage of following the principles of best practice over the procedures followed in automatic solutions. Finally, we highlight the principle of primacy of experience and professionalism over technical solutions.