A data-driven modal analysis of plasma oscillations in a stationary plasma thruster-100-like Hall thruster in the 1–120 kHz range is presented. Data are generated by a two-dimensional (axial-radial) hybrid particle-in-cell/fluid simulation code. While proper orthogonal decomposition is unable to successfully uncouple the different dynamics, higher order dynamic mode decomposition (HODMD) cleanly isolates the breathing and ion transit time (ITT) modes. Indeed, the computed HODMD components can be clustered into two distinct groups, enabling the separate reconstruction of the dynamics of the two oscillation modes. It is also shown that each plasma variable exhibits a different behavior in each cluster. The breathing oscillations have a global or standing-wave character for the plasma density, neutral density and ion current density inside the thruster, and an axially progressive-wave structure for the electron temperature and plasma potential in the near plume. The ITT mode shows a progressive-wave structure for all variables, except for the neutral density whose amplitude is negligible. This work serves as an illustration of the applicability of data-driven techniques like HODMD to the analysis of plasma flows relevant to electric propulsion.