Fraud Detection and Analysis System for Car Insurance Claim Using Random Forest Classifier

In recent years, commercial insurers have faced many cases of fraud in all types of claims. Fraud claims have been huge in amount and can cause serious problems. As a result, various organizations that are private or public, including the government, work to identify and prevent fraudulent activities. One of the most prevalent forms of fake insurance claims is related to the auto industry, which are often made through false accident claims. This project aims to solve this problem by developing a machine learning model that uses insurance claim datasets to detect and classify fraud and false claims from the legitimate ones. The project will use the Python PySpark library and the Random Forest Classifying algorithm to label and rank claims and compare their performance using metrics such as soft accuracy, precision, recall and confusion matrix.

I. Introduction

Insurance fraud is a serious problem that plagues the insurance industry and can cause substantial losses for insurance companies. It refers to the act of intentionally making false claims for money, often exceeding the actual amount of insurance that the company or other guarantor is obligated to pay. An increase in fraudulent activity is particularly common in the auto and insurance industries. Insurance fraud is an illegal act committed with the express purpose of obtaining financial gain. This is currently the most serious problem faced by many insurance companies around the world. Most of the time, one or more omissions in the investigation of misinformation have been identified as a major factor.