Electricity Theft detection Using Machine Learning Algorithms.
Abstract: With the rapid revolution in technology, the electricity has become the most essential asset of a country which can provide a reasonable momentum for its development. Electricity theft is a main challenging problem faced by the utility companies. Apart from technical loss (1L), most of the developed and developing countries are facing huge financial losses due to Non-technical loss (NTL) as well as electricity theft. The core objective of this thesis is to attempt to detect the electricity theft used by various machine learning algorithm. n this thesis we use some state of the art machine learning algorithms like SVM and RF classifier on a dataset that is originated from the State Grid Corporation of China. We have preprocessed the dataset using two methods: linear interpolation and data imputation. For each prepared data we have applied the two classifiers. We have achieved the best accuracy of 89.59% for linearly interpolated data with SVM and 88.88% for imputed data with RF to detect electricity theft.
|Class / Degree
Md. Sakib-Al-Emran  and Tanmoy Baidya 
|6th June, 2017
|20th December, 2018