This study deals with residential location modelling of Budapest, Hungary, using a neural network approach known as the self-organising map (SOM). After identifying various urban elements, the analysis focuses on the dynamics of two selected inner-city neighbourhoods: the middle parts of the districts VIII (Józsefváros) and IX (Ferencváros) respectively. These adjacent, but different, areas in the south-eastern part of the inner city have both received attention as subjects for substantial rehabilitation in recent decades. When the differentiation of micro-locations in the targeted areas was modelled using the SOM over a period of six yearly cross-sections, notable dynamics could be identified in terms of both actual changes in the fundamental quality, and expected house price escalations that are not related to fundamentals. In particular, the evidence reveals how housing market development is related to most localised processes of social and physical upgrading occurring in an urban setting.