Dynamic mode decomposition (DMD) is a data-driven spatiotemporal decomposition of complex systems, which downsizes them to a linear system. It can be used for the analysis, prediction, and control of dynamical systems. Specifically, DMD surrogate models have been successfully used for modeling crowd dynamics. Parametric DMD is an approach for modeling parameterized dynamical systems. This thesis considers a parameterized system from crowd dynamics. In particular, a bottleneck scenario is considered, where one room contains 1000 people who need to evacuate it through one door, which has adjustable width. The room is represented as a triangle mesh, where crowd densities for each mesh element are to be predicted over time. The problem’s parameterized nature consists of the adjustable door width, where multiple door width values are considered for a single surrogate model. This thesis analyzes, implements, and evaluates state-of-the-art parametric DMD methods for modeling this scenario. They successfully reconstruct the data and predict well for untested parameters. The gained insights serve as a foundation for further development of parametric DMD methods in the scope of crowd modeling.