The use of indicator kriging for analyzing prices in the real estate market

Abstract

The patterns and relations between real estate prices and the factors which shape them can be presented, among others, in the form of traditional statistical models, as well as by means of geostatistical methods. In the case of research involving the diagnosis and prediction of transaction prices, the key role is played by the spatial aspect, hence the particular significance of geostatistical methods using spatial information. The main goal of the conducted research is to determine the probability of the occurrence of a price in a given location in space by means of geostatistical simulation - indicator kriging. Indicator kriging does not use the entirety of information included in a dataset, and can, therefore, be useful in situations when the assumptions involving the spatial stationarity of the examined phenomenon are not fulfilled by an entire dataset, but are fulfilled by a certain part of the set. The maps of the probability with which a regionalized variable (price) takes on particular values, limited by arbitrarily selected cutoff values, were prepared by means of indicator kriging. An alternative approach to the preparation of price probability maps is the determination of the spatial distribution of areas in which, with the assumed probability, the value of the price falls within the predetermined ranges. The paper presents both the essence as well as a theoretical description of the geostatistical simulation of a transaction on the real estate market, as well as the results of an experiment involving the transaction prices of real properties located in the north-western part of the city of Olsztyn. The result of the research is a set of virtual information about the places in which the transactions have occurred and about the prices of real estate, constituting a reflection of the market processes which may take place in the near future.