Machine Learning is Employed to Detect Counterfeiting of Insurance Settlements

Abstract

An insurance company has been grappling with widespread fraud across various types of claims, prompting collaboration with government and organizations. This fraud issue poses serious financial risks due to significant fraudulent claims. The project's goal is to employ machine learning algorithms to analyze claim data, pinpointing fraud and inflated claims, particularly in severe cases like false accident claims in auto insurance. The project involves creating a model to assess and label claims, comparing machine learning algorithms using metrics like accuracy, precision, and recall via a confusion matrix. The PySpark Python library is used to build a fraud detection model. This industry- wide problem costs billions yearly, necessitating effective solutions to reduce fraud and unnecessary expenses. Key Words: learning, pyspark, crime identificaton