With the rise in data-driven organizations worldwide, large amounts of information are inevitably transferred, relocated and consigned on networks. These operations are done on the Internet and within organizational intranets and may often contain information of hi sensitivity. Thus, it becomes inevitable for companies and institutes to deploy network security protocols and strategies to counter cyber criminals and maintain the integrity and confidentiality of information and the availability infrastructure. This paper presents a structured, efficient and coherent approach to detect and prevent cyberattacks and breaches within a complex organisational network environment. It discusses the idea of ensemble learning (a sub-domain of machine learning) to create a stacking classifier that will predict the type of network attack the network infrastructure is most susceptible to. We achieve this by analyzing numerous network parameters. The model classifies the network probability of being susceptible to Denial of Service (DOS), Impersonation, and Man-in-the-Middle (MITM) attacks. It is classified as normal if none of the aforementioned attacks are detected as a threat. The model is trained using the OPCUA dataset and has proved to be a promising tool for network defensive evaluation.