© 2020 AISTI.In the insurance industry, fraud can be performed because of salesperson's bonuses and commissions. This problem reduces the reputation of the insurance industry. With the decrease of reputation, the trust of the customer to the industry decreases and thus, the acquisition of new customers becomes difficult. This study aims to estimate the probability of salesperson to fraud with historical data by using machine learning methods. The industry's reputation is maintained if fraud can be detected in advance.