Optimized Predictive Model For Insurance Claim Fraud Detection And Analysis Using Machine Learning

Abstract

Insurance Fraud Causes Significant Financial Losses And Is Difficult To Detect Due To Its Similarity To Legitimate Claims. The Growing Volume Of Data And Increasing Fraud Complexity Make Automated Detection Essential. Machine Learning Provides Effective Tools For Identifying Fraudulent Patterns In Insurance Claims.This Study Applies Logistic Regression (LR) And Support Vector Machine (SVM) For Insurance Fraud Detection. The Process Includes Data Preprocessing Such As Handling Missing Values, Feature Selection, Normalization, And Categorical Encoding. LR Estimates The Probability Of Fraud, While SVM Classifies Claims By Separating Fraudulent And Genuine Cases In A High-dimensional Space.A Comparative Analysis Using Accuracy, Precision, Recall, And F1-score Shows That Both Models Improve Fraud Detection And Reduce False Positives. The Study Highlights Their Respective Strengths And Limitations, Demonstrating How Machine Learning Enhances Fraud Detection, Reduces Losses, And Supports Better Decision-making In Insurance Systems.

Keywords