This study examines whether a statistically derived decision tree could serve as a means to improve U.S.A. Farm Service Agency lending decisions. The study is a substantial extension and reanalysis of an earlier work by Barney, Graves and Johnson, (1999). Results indicate that a decision tree could be a valuable tool for Farm Service Agency employees in their lending decisions. The decision tree provides as good or better predictive accuracy than neural networks and logistic regression models at reasonable cutoff levels of Type II to Type I costs of lending. The decision tree also meets the transparency criteria for Farm Service Agency purposes by providing logical, understandable rules for lending decisions.
Original Publication Information
Foster, Benjamin P., Jozef Zurada, and Douglas K. Barney. "Could Decision Trees Help Improve Farm Service Agency Lending Decisions?" 2010. Journal of Management Information and Decision Sciences (formerly Academy of Information and Management Sciences Journal) 13(1): 69-91.
Foster, Benjamin P.; Zurada, Jozef; and Barney, Douglas K., "Could decision trees help improve Farm Service Agency lending decisions?" (2010). Faculty Scholarship. 358.