You are given a dataset containing irregular trading settlement fees. How would you identify anomalies and clean the data for a cost-benefit write-up?