International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 03 | Mar 2024
p-ISSN: 2395-0072
www.irjet.net
Fraud Detection and Analysis for Insurance Claims Using Machine Learning 1Jaya Vani Vankara, 2V Seshadri Naidu, 3D Govardhan, 4V Vivek , 5P VNikhil 1Assistant Professor, Dept. of CSE, GITAM (Deemed to be University), Visakhapatnam, Andhra Pradesh, India. 2,3,4,5Student, GITAM (Deemed to be University), Visakhapatnam, Andhra Pradesh, India. ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Insurance claim fraud is a serious issue that the
mining and deep learning techniques, help identify patterns and anomalies in insurance claim data that may be signs of fraudulent conduct. These algorithms have the potential to increase the accuracy significantly. These advanced methods present the insurance sector with a viable way to improve fraud detection and lessen the effects of fraudulent claims. Insurance companies could save money, and consumers would feel safer if these algorithms significantly improved the accuracy and efficacy of fraud detection. These algorithms analyze various aspects of claim data, such as the kind of claim, policyholder information, and prior claim history, to spot abnormalities or questionable patterns. Insurance firms can create predictive models that use machine learning to assign a Fraud Probability Score (FPS) to each claim. This research primarily focuses on using machine learning to detect fraud with auto insurance.
insurance business faces. It costs insurance companies money and raises policyholders' rates. Machine learning has become a potential method for insurance claim fraud investigation and detection in recent years. Machine learning algorithms—such as data mining and deep learning techniques—have been effectively applied to identify trends and abnormalities in insurance claim data that point to fraudulent activity. These algorithms could significantly increase the precision and effectiveness of fraud detection, saving insurance firms money and giving consumers excellent safety. Machine learning algorithms can precisely analyze vast volumes of data and spot trends and abnormalities that can point to insurance claim fraud. These algorithms can look at various factors in claim data, including claim type, policyholder details, and past claim history, to identify anomalies or suspicious trends. With the help of machine learning, insurance companies will be able to build predictive models that will give each claim a Fraud Probability Score (FPS). In this project, we’re focusing on identifying auto insurance fraud using machine learning. An insurance agent should be able to investigate every case and determine if it’s real. But this not only takes time, but it’s also expensive. Hiring and financing the skilled labor needed to review every claim filed daily is impossible. This is where machine learning comes in. In this case, we will use one of the most widely used machine learning algorithms.
We have chosen one of the most popular machine learning methods to do this. These algorithms produced the highest accuracy in projected results and annual expenses, amounting to billions of dollars, proving their applicability to our dataset. Insurance fraud can take many forms in different insurance realms, and it can range in severity from small-scale claim embellishment to deliberate acts of destruction or harm. Auto insurance fraud is one of insurers' most significant and well-known problems. A claims agent should look at costs due to fraudulent claims, which highlights the significance of differentiating between genuine and fraudulent claims. Although a claims agent should look into each case separately, this is frequently an expensive, time-consuming, and inefficient procedure. Examining all of the many claims that are filed every day would be very impossible. To detect and mitigate fraudulent claims, machine learning offers a practical, quick, and economical solution.
Key Words: Fraud Insurance, XGBoost, Artificial Neural Network, Random Forest, Logistic Regression, Decision Tree, SVC. 1. INTRODUCTION We have discovered a significant issue with insurance fraud in this project. Claims filed to deceive an insurance company are known as false coverage claims. Since the beginning of the insurance sector, there has been a persistent problem with insurance fraud, with a significant portion of received claims being fake. Insurance firms suffer financial losses from fraudulent claims, and policyholders pay higher premiums. Machine learning algorithms, which use data
© 2024, IRJET
|
Impact Factor value: 8.226
2. Literature Survey [1] T. Badriyah, Lailul Rahmanian I. Syarif, titled 'Nearest Neighbour and Statistics Method based for Detecting Fraud m Auto Insurance" provides an overview of the nearest
|
ISO 9001:2008 Certified Journal
|
Page 630