Application of Mixed-Effect Regression Models for Compiling a Land Value Map

Cartographic representation of land value can be a result of regression analysis taking into consideration the spatial distribution of factors affecting location values. Classical regression models do not always fulfil their role due to a hierarchical structure of market data. The paper proposes the application of mixed-effect models, including two levels of an analysis, with the function of the property in the local zoning plan as the grouping variable. The study was carried out in Olsztyn, a city located in the north-eastern part of Poland. Physical, legal and location features were used as explanatory variables. A model containing fixed and random effects made it possible to develop a theoretical distribution of the city land value depending on the function fulfilled by the property. This provided a basis to compile a land value map, which reflects both the distribution of prices and the spatial factors affecting them.