Title Intelligent System for Detection of Abnormalities and Probable Fraud by Metered Customers
Abdul Rahim Ahmad, Fariq Izwan Ismail, Abdul Malik Mohamad · 2007
This paper describes an on going project in Tenaga Nasional Berhad, (TNB) Malaysia to create an intelligent system to detect fraud by customers of SESB, a subsidiary of TNB in Sabah, East Malaysia. It presents a methodology to obtain a list of abnormal users from the customer database using a popular and recent intelligent method, the support vector machine (SVM). Instead of spending a lot of money on inspection campaign on all customers, a list of likely abnormal (fraud) customers will be generated and checked. First, an SVM model is constructed using samples of verified customer list. Then, the model is used for the detection of fraud customers from the 400,000 customers.