This article develops a mathematical model for residential location choice in which previous experience is considered via a dynamic learning process and each agent evaluates location decisions according to their utility. Agents' idiosyncratic behavior is modeled, assuming a stochastic willingness to pay based on dwelling and neighborhood attributes and households' socioeconomics. Additionally, the willingness to pay is affected by the agent's experience of the urban context in previous periods. Numerical examples are given, and simulations are conducted using linear bid functions. Additionally, concepts of urban dynamics are used in the long-term, assuming quasi-equilibrium in the bids considering the externalities of urban configurations. The results are compared with a static outcome corresponding to the equilibrium model of land use. The comparison is performed using indices of urban segregation and both short- and long-term configurations. The studied dynamics show that static modeling does not explain all the features associated with such a configuration of the urban system (in both short and long term).