A machine learning specialist (ML) working for an investment firm is building a credit risk model to identify potentially risky loans. The training data includes 10,000 cases collected from past customers. Upon inspection, the specialist found that 95% of the cases consisted of fully repaid loans, and only 5% defaulted loans. The company wants to minimize the number of loan rejections from non-defaulters. Which method can the specialist implement to satisfy the requirements with the LEAST operational overhead?