Fraud detection in customers’ electricity consumption in Nigeria using machine learning approach

Authors

  • Sulaiman Olaniyi Abdulsalam Kwara State University, Malete Author
  • Micheal Olaolu Arowolo Landmark University, Omu-Aran Author
  • Ronke Babatunde Kwara State University, Malete Author
  • Musa Raji Kwara State University, Malete Author
  • Shakirat Oluwatosin Haroon Sulyman University of Ilorin Author

Abstract

Electricity theft is estimated to have cost Nigeria billions of Naira over the years. Electric utilities use data analytics to discover unusual consumption patterns and possible fraud in order to prevent electricity theft. This work uses data analysis to detect electricity theft, as well as a measure that uses this threat model to compare and evaluate anomaly detectors. This study employs machine learning algorithms to categorize fraud detection in customers’ electricity use, as data mining techniques has helped multiple companies and sectors better their various types of technology. Support Vector Machine (SVM) and C4.5 Decision Tree classification algorithms were used to detect fraud using consumer electricity use data. The accuracy of SVM and C4.5 was 63.4 percent and 65.9%, respectively. As a result, the Map-Reduced-ANOVA with SVM attained an accuracy of 77.5 %.

Downloads

Published

2022-02-17

Issue

Vol. 2 No. 1 (2021)

Section

Computer & Information Sciences

How to Cite

Abdulsalam, S. O., Arowolo, M. O., Babatunde, R., Raji, M., & Sulyman, S. O. H. (2022). Fraud detection in customers’ electricity consumption in Nigeria using machine learning approach. Technoscience Journal for Community Development in Africa, 2(1), 81–91. http://kwasu.site/index.php/technoscience/article/view/30