Data mining in a power company customer database

Abstract The paper deals with a successful attempt to extract hidden knowledge by means of intelligent treatment of a large amount of data stored in a power company database, which is used to support large customers billing activity. These databases usually record a huge number of tiny pieces of information, mainly related to administrative facts, contractual data, billing procedures and consumption recordings, but they do not correlate these data. The methodology adopted is based on a mixture of techniques taken from artificial intelligence, such as: artificial neural networks and fuzzy logic, which, recently, have been widely adopted to solve power system operational problems. It was noted that the proper use of these techniques allows management of a large amount of data, introducing both added values in old fashion databases and a new approach on how to consider the existing information.