ملخص الدراسة:

Non-technical losses are the problem that concerns all electricity utilities around the world. Data mining algorithms are used to minimize the problem and detect the most proper subscribers for theft; making them the target to the inspection staff periodic detection for the electric meters. Taking into consideration the specialty of Palestine as the case study, the two major non-technical losses found is tampering with meters or having illegal connection at the network. The second case made the traditional data mining technique that depends on the consumption history of the subscriber little weak. Also the detected frauds admitted that their consumption after fraud was double folds the estimated from the utility; that fact side by side with the fact that most of violators also distribute electricity; makes the consumption record only little weak for detection. The idea is a classified data mining technique that takes into consideration the general information about the subscribers as influencing factor. When talking about theft we are talking about illegal electrical consumption despite the way of doing it. Researchers around the world use the consumption patterns to detect fraud. Locally, people do frauds without having the meter from the beginning that yields to missing patterns records for many cases. The idea is to find the reasons that drive people to theft; that I did and more research can be done also on this cases; and then constructing a database for all the influencing information for theft about the subscriber from the first time apply for the meter. For the cases that have no meter at all and of course the utilities knows nothing about them, the system will help a lot as it will classify any input information to it with either YES or NO fraud. In this work the RapidMiner software was used to compare the different techniques for classifications: the decision tree with its different types and the Naive Bayes classifier.

توثيق المرجعي (APA)

Nayfeh, Rana O,& Abu-Issa, Abdallatif (2016). Using Classification Data Mining Technique to predict the Non-Technical Losses in Power Systems in Palestine. INTERNATIONAL CONFERENCE ON SMART CITIES SOLUTIONS, The Islamic University of GAZA. 28254