System identification (SI) is an important subfield of control engineering. In structural dynamics, SI methods allow for the extraction of modal parameters, thus enabling the dynamic characterisation of structures. Recently, an efficient frequency domain identification method, the Loewner Framework (LF), has been extended to modal parameter extraction. The LF is based on tangential interpolation, and as such, data partitioning is fundamental for its operation. Traditionally, real-valued data is used for the fitting process. This means that complex-valued transfer function measurements are forced into the real domain prior to fitting, resulting in a computational penalty. This work proposes to validate the use of complex-valued data partitioning within structural dynamics to increase the computational efficiency of LF. For this aim, a numerical model of a mass-spring-damper system is modelled; then, its dynamic response is simulated in MATLAB 2024b, while considering different levels of noise. The results show that the complex-valued data sampling identification results perfectly match those using real-valued data, theoretically ought to be better for solving the system realisation problem via tangential interpolation.