A Benefit Optimization to Motor Insurance Claims Fraud Detection using Machine Learning
Do a literature review for a project on using machine learning and data science for Fraud Detection on Motor Insurance Claims. The aim of the project is to predict which motor insurance claims are fraudulent or suspected to be fraudulent. This is determined by minimizing costs.
A breakdown of the project:
– Take the dataset of unlabeled motor insurance claim data and use unsupervised techniques for Anomaly detection (mainly Isolation Forest)
– Use that result, oversample using SMOTE and then use classification (mainly XGBoost and Logistic Regression) to obtain a confusion matrix.
– Then that cost is minimized ( Benefit Optimization approach to the Evaluation of Classification Algorithms attached) by minimizing the cost of the False Positives and False Negatives
