Access control is vital for protecting organizational resources, with Role-based access control (RBAC) offering a widely adopted framework for managing permissions. However, static role assignments in traditional RBAC systems often become misaligned with evolving organizational structures and user behaviors, leading to inefficiencies and potential security risks. This paper explores a predictive enhancement to RBAC policies by analyzing historical access logs using Hierarchical Clustering (HCL) techniques. The proposed approach uncovers behavioral access patterns to support data-driven refinement of role assignments. By incorporating behavioral clustering into access control workflows, the method helps align permissions with observed usage trends and may reduce excessive privilege assignments. Evaluation on a real-world dataset demonstrates that the model adapts roles based on access behavior, offering a step toward more responsive and behavior-aware access governance.