In this thesis self -survey results were revealed relating to the application of the artificial neural networks used for specifying market models of business premises’ rental rates taken from the selected urban commune’s resource. The research procedure consists in training selected networks with data from the local real estate market, and transforming detected relationships – through the agency of established models – to estimating the potential rates of the market city council accommodation rentals. On this basis the market adequacy degree of the actual rates of commune’s premises rentals was estimated for the commercial, service, storage and public functions. Empirical researches were carried out on the local real estate market of the city of Olsztyn. In order to describe the phenomenon of self-moulding of the market business premises rental, one-way three-level and radial elementary function networks have been chosen. Analyses have revealed the relatively low degree of convergence between the actual city council’s accomodation rentals and adequate rentals potentially determined by the market. This degree was strongly diverse depending on the type of business being run in the premises, and the evaluation of its influence on the society and the urban economy. The research methodology used, as well as received outcomes, can serve to streamlining commune’s premises resourses management, elicitation of rental policy including.

55-67