The construction sector is linked to the general development of a country. There is a lot of data scattered and not properly explored in relation to the buildings constructed. However, if these scattered data on the behaviour of the real estate market are organized, combined with knowledge of civil engineering, this merger of information can mitigate some evaluation problems, especially those that are overvalued for unknown or dubious reasons. Thus, there is a need for models capable of working with limited data to analyse the causal relationships between explanatory variables and sales prices and, from there, predict property values. The purpose of this article is the innovative use of simple building inspection strategies to predict the market price for residential apartments. For this, 19 samples of residential apartments are used in the city of Niterói, Rio de Janeiro, Brazil, in February 2021. The methodology uses the results of the survey of civil engineering and converts them into heuristic terms predicting the price of the property. With this, the imprecision, uncertainty, and subjectivity of human expression combined with the knowledge of civil engineering result in a plausible solution and easy application in the market. Finally, the use of fuzzy logic in the evaluation of properties is an adequate unconventional method, in addition to avoiding repetition in regression coefficients in binary logic. To check the reliability of the method, the comparison between the market values of the samples and the values predicted by the fuzzy logic is used. The result according to the mean absolute percentage error (MAPE) can be interpreted as a good result (7%).

One of the most challenging tasks in modelling of house pricing is to take into account the location factors. Geographically Weighted Regression (GWR) as a local regression model is an extremely effective instrument for spatial data analysis. The aim of the study is to model the relationships between a residential real estate price (per sq.m) and both building and location characteristics for Gomel using GWR. The data of the Belarus’ National Cadastral Agency on real estate transactions (apartments) in Gomel in 2019 are used as initial. The global Moran I index has been used to estimate a spatial autocorrelation of the dependent variable (price per square meter of residential real estate). Several factors having the impact on the apartment sale prices have been determined. Independent variables having been used in analysis can be divided into building characteristics and spatial characteristics. The building characteristics section includes the number of rooms within the property, property area (square meters), building age, number of floors in the building, floor of the property. The spatial characteristics group contains proximity to city centre, recreation areas, supermarkets, bus stops, healthcare and educational facilities. A regression model of housing price in Gomel has been developed. Mapping variable regression coefficients allows exploring spatial features of the impact of the different explanatory variables on the property price. Geographically weighted regression modelling has revealed the pricing peculiarities inherent for certain areas of the city.