Power theft detection using random forest algorithm and compared with K-Nn algorithm

• Published in: AIP conference proceedings Vol. 2853; no. 1
• Main Authors: Reddy, K. Bhupal, Malathi, K., Priya, M. Vishnu
• Format: Journal Article Conference Proceeding
• Language: English
• Published: Melville American Institute of Physics 07.05.2024
• Subjects: Algorithms; Power consumption; Sensitivity; Statistical analysis; Theft
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0198770

The aim of the study is to diminish the power loss with reference to sensitivity and precision in power transmissions and consumptions. As power theft is considered as non-technical loss, it is hard to track the total transmission. So the Random forest algorithm is proposed over the K-NN Algorithm to compare accuracy in power theft detection. Materials and Methods: There are two groups in this study each with a sample size of 21400 per group. Analysis is done with the pretest power 0.8. Results: The mean value for accuracy in the Random Forest algorithm is 90.6215 which is high when compared to K-NN algorithm whose accuracy is 82.2200. It has an insignificant value of 0.07 (p>0.05). These values are evaluated using SPSS for statistical analysis. Conclusion: This analysis shows that the novel Random Forest algorithm has more reliable accuracy and sensitivity of power theft detection compared to the K-NN algorithm.