Leveraging random forest algorithm for proactive detection of information breaches in the Nigerian oil and gas sector

The Nigerian oil and gas sector's dependence on information and communication technologies makes it vulnerable to various cybersecurity threats, such as data breaches that can disrupt operations and cause significant financial losses. Despite implementing several security measures like firewalls, authentication systems, and intrusion detection systems, they struggle to keep up with the rapidly changing threat landscape. However, machine learning techniques offer potential for strengthening security in the oil and gas industry. Therefore, this study collected data on cyberattacks from relevant organizations, analyzed, cleaned, preprocessed, and split the dataset into training and testing sets using an 80/20 ratio with Python tools on Google Colab. The research identified common attack vectors, which were used to develop a machine learning model aimed at detecting and mitigating cyber incidents. The initial phase of model development faced challenges due to imbalanced datasets, leading to biased results. By applying data balancing techniques like SMOTE, the model's accuracy increased from 75% to 92%. Addressing the class imbalance also enhanced other evaluation metrics like recall, precision, and F1 score, demonstrating the model's potential for more accurate threat detection and prevention. The study concludes that integrating machine learning with cybersecurity can improve proactive security measures for infrastructure in the oil and gas sector through early warning systems with minimal false positives. This highlights the importance of advancing cybersecurity practices in Nigeria's oil and gas industry by leveraging machine learning to adapt to and counteract emerging threats effectively.