In this study, a data-driven subspace system identification approach is proposed for modeling guided wave propagation in plate media. In the data-driven approach, the subspace system identification estimates a mathematical model fitted to experimentally measured data, but the black-box model identified captures the dynamics of wave propagation. To demonstrate the versatility of the black-box model, wave motions in various shapes of aluminum plates are investigated in the study. In addition, a waveform predictor and temperature change indicator are proposed as applications of the black-box models, to further promote the modeling approach to guided wave propagation.