This paper illustrates how fuzzy logic can be helpful for constructing event-type variables in operational risk management. Even when the available databases cannot be considered "native" fuzzy, we show that modeling them according to fuzzy intervals is useful for two reasons. First, it allows more information to be taken into account, and exploited, and second, predictive models applied to this kind of data perform comparatively well. The paper shows how to organize event type variables into macro classes using fuzzy variables, and also shows how such variables can improve predictive performance.
Back to Top