Control list generation to reveal fraudulent customers for electricity provider companies

This paper presents a procedure that generates control lists for electricity provider companies to reveal fraudulent customers with the highest possible efficiency. Based on the database of Customer Registration System, essential features are selected and other secondary (derived) properties are introduced to build up an abstract space for the algorithm of control list generation. Control list generation algorithm uses training data from the Customer Registration System to set up a fuzzy system with subtractive clustering. The algorithm has been validated at Hungarian Electricity Provider ELMU-ÉMÁSZ Company Group.