Developing a typology of heterogeneous fishing practices through the use of métier analysis is a useful step in understanding the dynamics of fishing fleets and enabling effective implementation of management outcomes. We develop a non-hierarchical clustering framework to quantitatively categorize individual fishing events to a particular métier based on corresponding catch composition, gear configuration, and spatial and temporal references. Our clustering framework has several innovations over predecessors including: (i) introducing alternative methods for encoding and transforming fisheries data; (ii) variable (feature) selection methods; (iii) complementary metrics and methods for internal métier validation; and (iv) use of a network science method to model and analyze fishing practices. To demonstrate applicability, we apply this framework to the Australian Eastern Tuna and Billfish Fishery (ETBF), a multispecies pelagic longline fishery with a diversity of fishing practices. We identified a total of seven stable métiers within the ETBF. While each métier was characterized by a predominant target species, they were differentiated more by seasonal and temporal references (e.g., time of set, month, latitude) than gear configuration (e.g., hooks per basket) or target species. By collapsing a large amount of high-dimensional operational data into a relatively uniform and limited number of components, decision-makers can more easily evaluate the likely consequences of management and design policies that target a particular métier.