Using data mining to detect anomalous producer behavior: An analysis of soybean production and the federal crop insurance program

Research regarding anomalous producer and county soybean production was conducted on the United States Department of Agriculture/Risk Management Agency Book of Business for reinsurance years 1994 to 2001. The purpose of the analysis was to assist RMA in the identification of anomalous producers and analyze the relationship between anomalous producers and anomalous counties by developing a data mining pattern detection model. Identification of producers and counties as anomalous is not prima facie evidence of fraud. However, the outlier detection of anomalous producers and anomalous counties has raised some interesting questions because both sets of outliers cluster at or near Land Resource Region boundaries.