Correlation Analysis in the Process of Weighting Real Property Attributes

The key goal of the market analysis carried out for property valuation purposes is to select pricing attributes of real properties and to assign so-called weights to them, which would illustrate their influence on the prices in a given market (Regulation 2004). Correlation analysis is a very useful tool in this respect. However, the fact that it is limited to the use of classical Pearson’s linear correlation coefficients is too much of a simplification due to the frequently occurring heterogeneity of the real estate market.

The research paper proposes to use the possibilities offered by the broadly understood correlation analysis which, among linear correlations, takes into account not only Pearson’s correlation but also rank correlations. At the same time, analyses of various non-linear correlations are being carried out where linear relationships are not reliable enough. The aim of this research is the simultaneous verification of the existence of various types of correlative relationships, taking into account the nature of random variables that the analyses relate to, and the nature of relationships between them. This approach makes it possible to adjust the weighting of market attributes, the values of which are improved in a few subsequent steps, in order to eventually approach the optimal result as close as possible.