Campus wireless LANs (WLANs) are complex systems with hundreds of access points (APs) and thousands of users. To analyze the performance of wireless networking protocols, researchers need to construct simulations and testbed experiments that reproduce the characteristics of these networks. However, the generation of realistic models and benchmarks is challenging and there is only a limited set of models of roaming and access based on real measurement data. We employed graph theory, modeled the roaming activity as a graph and measured its degree of connectivity. The Negative Binomial distribution models well the degree of connectivity. Furthermore, we analyzed the evolution of the roaming activity in the spatial and temporal domain and its impact on the degree of connectivity of the graph.