Fraud detection in electrical energy consumers using rough sets

Rough sets is an emergent technique of Soft Computing that have been used in many knowledge discovery in database applications. This work describes an application of rough sets in the fraud detection of electrical energy consumers. From an information system, rough sets concept of reduct was used to reduce the number of conditional attributes and the minimal decision algorithm (MDA) was used to reduce some values of conditional attributes. The reduced information system derives a set of rules that reaches consumers behavior, allowing the classi cation rule system to predict many fraud consumers pro les. Rough sets proves that it is a powerful technique with application in many systems based in data.