Prediction of Health Insurance Claims Using Logistic Regression and XGBoost Methods

Abstract

Fraud is one of the crimes in the financial industry. Health insurance claims are a kind of fraud in insurance. This research examines health insurance claims to predict fraud. The claims would be denied if there were any suspicions of fraud. Extreme Gradient Boosting (XGBoost) and Logistic Regression are the approaches employed in this research. The most recent method in machine learning, XGBoost, is an improvement over the classic statistical method, Logistic Regression, in terms of classification performance. There are nine independent variables that influence fraud in health insurance claims, according to the findings of the Logistic regression model. The comparison between Logistic regression and the XGBoost model showed that the XGBoost model is better than the Logistic model. Based on the result of accuracy, precision, and recall values on both models.