The model reduction technique exemplifies the approach of representing solutions within higher-dimensional data through a lower-order spatial framework. This methodology serves to alleviate complexity by enabling the representation of solutions through a reduced set of modes. In this context, we shall endeavor to construct a lower-order representation of a scramjet system employing both linear and non-linear techniques. The linear reduced order model (ROM) utilized in our research encompasses dynamic mode decomposition (DMD) and Proper Orthogonal Decomposition (POD). The non-linear ROM, on the other hand, is predicated upon an artificial neural network (ANN). This ROM is poised to facilitate the analysis of flow characteristics, optimization of geometrical configurations, and the design of a scramjet controller. Within the scope of this study, we will conduct a comparative assessment of the performance of ROMs obtained through the linear reduced order model and the ANN-based model, leveraging Computational Fluid Dynamics (CFD) data. Additionally, we will assess and compare the computational efficiency of these distinct methodologies.