Electricity Theft Detection Using K-means Clustering in Electricity Information System

Electricity is the most important component in life to support daily activities. Based on the obligations of consumers, who will hereinafter be referred to as customers, customers are obliged, among other things, to utilize electric power according to its intended purpose. In reality, up to now there are still many customers who commit violations regarding the use of electricity. This research aims to detect customers who commit electricity theft based on a collection of electricity information data obtained from Automatic Meter Reading (AMR). The method used is the k-means clustering method. First, preprocess the data for each customer. Then an analysis of the data set profile is carried out using k-means. The customer whose data set profile has the smallest proportion in the cluster can be said to have committed electricity theft. The expected research result is that it can detect customers who commit electricity theft. So, this can reduce losses experienced by state electricity companies.