In order to predict radio propagation behaviour within a context-aware scenario and to make better decisions on the deployment of antennas and sensors for particular purposes, the state of the art proposes plenty of simulation techniques based on deterministic methods like Ray Tracing and Ray Launching, combining Geometric Optics and Uniform Theory of Diffraction. In this context, simulations' quality largely depends on the number of reflections, providing sufficient angular resolution in the ray launching process. Despite of that, the computational cost of simulations in high definition (HD) prevents their use in complex environments and their low definition (LD) counterparts are used. In this article, we study the impact of simulation's parameters in order to discover which is the best possible configuration given a particular scenario. More concretely, we study the number of reflections parameter and its impact on post-processing techniques such as hybrid methods based on Collaborative Filtering techniques.