One of the most important housing research issues is the interplay of socio-cultural factors (lifestyles) and of objective constraints (increasingly important in tense metropolitan housing markets) in shaping households' housing preferences. This paper presents a modelling application for interview/survey data analysis on home ownership within metropolitan areas. The goal of the application is to develop an exploratory methodological protocol to analyse the interplay of lifestyles and objective constraints in home ownership decisions, as well as the spatial inscription of residential trajectories and of mobility behaviours. Its final result is the identification of coherent socio-geographic profiles capable of segmenting new home owners within the study areas of Provence and of the French Riviera. Socio-geographic profiles are identified and characterized by a new five step clustering protocol using Bayesian Networks (BN). BN modelling can integrate both qualitative and quantitative data, and characterise profiles in a probabilistic framework. Socially appreciated suburban and as well as urban lifestyles are present in Provence and on the French Riviera, shaping particularly similar socio-geographic profiles. Peculiarities of the geographical configuration of the two metropolitan areas and of their housing stocks can explain the residual differences among the corresponding profiles.