Analysis of factors affecting housing sale prices in Trabzon using the hedonic price model and quantile regression

Creative Commons License

TEMEL B. A.

Journal of Innovative Engineering and Natural Science, cilt.6, sa.2, ss.377-398, 2026 (Scopus, TRDizin)

Özet

This study aims to identify the structural, amenity-related, and locational factors affecting housing sale prices in Ortahisar, Trabzon. The research was conducted using 1,476 observations obtained from online housing listings for February 2026. First, a hedonic price model was developed to determine the main variables affecting housing prices. Then, quantile regression analysis was applied to examine whether these effects varied across different price segments. A Random Forest-based SHAP analysis was also used to support the findings within a more flexible modelling framework. The findings show that net area (m²), building age, floor level, elevator availability, being located within a residential complex, heating system, parking type, and neighborhood effects play significant roles in housing prices. In particular, net area (m²) was found to be one of the strongest determinants of price, whereas building age had a negative effect. The quantile regression results revealed that price determinants do not operate with the same strength across all market segments; variables such as net area (m²) and elevator availability produced stronger effects in upper price segments, while the negative effect of building age became more pronounced in lower segments. The SHAP-based complementary analysis largely confirmed these findings by making the relative importance, effect direction, and possible nonlinear patterns of the variables more visible. Overall, price formation in the Ortahisar housing market was found to be non-homogeneous, multidimensional, and differentiated across price segments. The results also offer practical implications for construction firms, housing developers, investors, and valuation professionals by showing how structural, amenity-related, and locational housing attributes generate varying levels of market value across price segments.