Research into social network systems has observed a phenomenon, named tolerance, through simulations. Tolerance is defined as agent's willingness to maintain an unrewarding connection. We have presented a mathematical model of tolerance to predict the n-tolerance of an agent. We use a real world dataset in the experiments and adjust variables and procedure to achieve desirable results. This work has shown that tolerance is not just a side-effect of the decision rule chosen for the system, nor is it a purely sociological theory, but an observable trait of dynamic social networks in real-world situations.