A new age of distributed and decentralized computing architectures has emerged in recent years due to the rise of edge computing. However, despite its potential, edge computing faces a variety of challenges, and one critical issue lies in the optimization of resource utilization due to the diverse nature of edge environments. Although resource allocation regards a topic heavily discussed in the area of edge computing, traditional approaches cannot manage to address efficiently this problem due to the heterogeneous nature of such environments. Motivated by this, in this paper, a novel dynamic resource allocation strategy for edge computing infrastructures is proposed, capable of identifying contextual information and causal relationships between the factors that affect an edge computing system and encapsulate them into the framework in order to perform informed adaptation decisions. The proposed framework is evaluated extensively in a simulated environment and the results show that it manages to optimize resource allocation and enhance the overall performance of the edge computing environment significantly when utilized.