Widely recognized since the beginning of air travel as a major issue, noise reduction remains nowadays a pressing concern for all stakeholders in the aviation industry. While aeroengine compressors, specially at the approach phase, have been historically identified as a leading source of noise, most of the research has been conducted on compressors of the axial type. However, radial compressors are found in a wide array of applications: smaller business jets, helicopters, unmanned aerial vehicles (UAVs), auxiliary power units (APUs), turbochargers for reciprocating engines, etc. Owing to their geometrical particularities, radial compressors feature flow patterns that differ from their axial counterparts, leading to different acoustic performance but also opening the door for different optimization approaches. Yet, classical modal decomposition techniques focused on duct propagation may fail to reveal the complex interactions between geometry and flow features that act as noise sources. In this paper we apply, in addition to the classical approach, a data-driven Dynamic Mode Decomposition (DMD) to pressure data coming from a Detached Eddy Simulation (DES), in which we have experimentally validated the correct reproduction of the modal behaviour of the compressor, thus obtaining in-depth details of the link between flow phenomena and noise generation and transmission across the inlet and outlet ducts.