Anomaly Detection Frameworks for Cybersecurity in Nigerian Fintech Startups Using Unsupervised Machine Learning Techniques

This research establishes an unsupervised machine learning-based anomaly detector that will enhance cybersecurity among Nigerian fintech startups where high rates of digital growth, volumes of transactions, and critical shortages of labeled fraud data render traditional rulebased and supervised fraud detection systems ineffective. The need to have adaptive and scalable security systems capable of identifying emerging and novel fraud patterns and keeping computational efficiency and operational feasibility in resource-constrained fintech contexts drives the study. The study employs the real-life experience of actual transactions with a financial technology application in Nigeria to model standard transactional behavior and identify abnormalities as a deviation of the standard. It does it in a systematic process, which includes data preprocessing, feature normalization and behavioral analysis.