ENHANCING CYBERSECURITY IN ANAMBRA STATE, NIGERIA, THROUGH A NOVEL FRAMEWORK FOR DETECTING INSIDER THREATS IN SMALL AND MEDIUM ENTERPRISES USING MACHINE LEARNING

Abstract

White collar criminals are a serious security issue to the organization of cybersecurity, especially in small and medium enterprise (SME) where funds to build advanced defenses are unavailable. The paper will build a new machine learning model to be used to identify insider threats in SMEs in Anambra State, Nigeria, and deal with the most common trends like the inappropriate access to data and suspicious user activities. The main idea is to detect the patterns of the threats with the help of the empirical data, create a diagnosis model based on ML along with the application of the Random Forest algorithm, analyze the model with the help of the real-life examples, and recommend scaling solutions to further deploy it on a larger scale to Nigeria. The study is a mixed-method study, undertaken in commercial centers such as Onitsha, Nnewi and Awka. This study will use 50 SMEs as the target population and stratified random sampling will employ sampling proportions of 50 out of the target population of 50 SMEs. Structured questionnaires of 250 employees and 60 managers, semi structured interviews with 25 owners and anonymized system logs are to be used as primary data collection. Random Forest model, which was trained on features such as the frequency of the logging in as well as access to files, was able to validate at 88% and 90 percent. The demographics will show that the background of the participants is mainly male (65%), 25-45 (70%), secondary education (55%) as the workforce in the SME sector in Anambra. The qualitative data provide such contextual variables as insufficient cybersecurity awareness and financial limitations contributing to the threat. The framework provides a lightweight and cheap tool that can use minuscule resources to facilitate cybersecurity efforts in environments that are resource constrained which helps in building resilience to cybersecurity in constrained environments. This study, offering behavioral analytics with local data-driven data, helps to bridge the gaps in the context-specific studies facilitating the policy recommendations on the improved SME protections and the national cybersecurity approaches.