The division of a national territory is a mandatory process to analyse socio-economic dynamics. Commuting is then an important dimension to build such classification and weighted network analysis is adapted to study this phenomenon. We present in this paper a procedure to help users to identify hierarchical partitions of cities that capture commuters flows density. We enforce our method on a network which represents commuting in France (based on the 1999 national census). Our approach is based on a common technique improved by visual tools: highlight dense areas using a strength metric and extract clusters at different levels using the variation of a quality measure function.