Managing user mobility and allocating resources optimally in distributed computing infrastructures have become more difficult due to the Internet of Things (IoT) explosive growth. Traditional cloud architectures often face latency and bandwidth limitations, making them less suitable for real-time and dynamic mobile applications. To solve these problems, this study proposes a machine learning-based approach to enhance performance in fog computing environments. The iFogSim three clustering algorithms are evaluated using a simulation framework: Kmeans, Self-Organizing Maps (SOM) and Dynamic Distributed Clustering (DDC). The proposed models aim to improve the categorization of mobile users, reduce energy usage, and increase efficiency in resource utilization at the network edge. According to experimental findings, SOM routinely performs better than K-means, particularly when integrated with advanced resource management mechanisms. Furthermore, both SOM and K-means demonstrate significant improvements over DDC without resource control. These findings highlight the potential of machine learning-driven strategies to provide scalable, adaptive, and energy-efficient solutions for mobility and resource management in mobile fog environments.