Thanuj Kumar S, Utsav Deep, Syed Shoiab, Syed Atif, Tejas Bhatnagar and T. Ramesh, Department of Computer Science and Engineering, Presidency University, Bengaluru, India.
International Journal of Advanced Information and Communication Technology
Received On : 10 Nov 2020
Revised On : 31 Nov 2020
Accepted On : 22 Dec 2020
Published On : 05 January 2021
Volume 08, Issue 01
Pages : 001-004
Abstract
Every year, the insurance industry losing billions of dollars due to fraud. The act when a person makes fake insurance claims to gain benefits, compensation & other advantages to which they are not entitled is known as Insurance Fraud. Nowadays insurance fraud detection is becoming a tedious problem for insurance companies to deal with as they need more investment and workforces to keep track of every transaction. In this paper, we are focusing on the major issue faced by insurance companies that is insurance fraud. we use the machine learning technique to detect insurance fraud based on the transactional data given by the insurance company. We build predictive models and compare their performance by calculation of confusion matrix then it is evaluated on various performance measuring parameters like accuracy, precision, recall, F1 score, and on AUC curve. SVM (Support Vector Machine) and XG Boost (Extreme Gradient Boosting) are the machine learning algorithms used. After model evaluation, we select the best model for prediction.
Keywords
Fraud Detection, Insurance Fraud, Machine Learning, Performance.
Cite this article
Thanuj Kumar S, Utsav Deep, Syed Shoiab, Syed Atif, Tejas Bhatnagar and T. Ramesh, “Insurance Fraud Detection Using Machine Learning, ” INTERNATIONAL JOURNAL OF ADVANCED INFORMATION AND COMMUNICATION TECHNOLOGY, pp. 001-004, January. 2021.
Copyright
© 2020 Thanuj Kumar S, Utsav Deep, Syed Shoiab, Syed Atif, Tejas Bhatnagar and T. Ramesh. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.