Process mining is crucial in improving processes by offering understandable process models that can be examined and improved. However, making sense of the results from process mining for human behavior analysis, which can be complex, requires transforming detailed, low-level tasks into simpler, high-level ones through event log abstraction. The main aim is to make complex human behavior models more straightforward. One unique aspect of this study introduces a dynamic time threshold method. Unlike traditional event log studies that rely on fixed time limits, this study uses a flexible approach where time limits adjust automatically based on each activity. This innovation eliminates the need for a predetermined time limit, making it more useful in real-life situations. The research findings highlight how understanding the subtleties of human behavior is crucial for optimizing IoT systems and establishing process mining as a valuable tool for analyzing human behavior in the context of IoT. Utilizing event log abstraction significantly improves the model’s clarity, and its effectiveness is evaluated using fitness and precision measurements. The results suggest that the models can provide practical insights, enhancing people’s experiences with IoT systems.