Electricity Customers Segmentation: A Mechanism to Curb Non-technical Losses in Power
Hadiza Ali Umar,Ibrahim A. Lawal
Keywords: Non-technical loss, Gaussian Mixture Model, Customer Segmentation
Abstract
Power utilities globally have a huge task of enhancing their services to enable them to compete in emerging markets. However, power utilities in Nigeria cannot meet their obligations due to challenges of revenue deficits that ensue due to collection losses known as Non-Technical losses (NTL). Utilities and government agencies must continue to place a high premium on assessing the intensity of NTLs to boost revenue generation and enhance the dependability of the power value chain. Using machine learning (ML) techniques, researchers can leverage the massive amounts of data produced by power companies to determine customer trends in energy payment and usage. Segmenting electricity customers with similar characteristics into behaviourally comparable groups is one use for this technique. This study groups customers into segments based on tariff plans and bill payments using the k-means, DBSCAN, and expectation maximisation clustering algorithms. The goal is to segment the customer base identify outliers and.