Fraudulent Claims Prediction
June, 2020
Predictive Analytics
Client
- Name:Confidential
- Services:Health Insurance
Objective
We were tasked with building a system to generate predictions of fraudulent claims data using a deep feature engineering process and predictive models. This allowed us to detect the nature of the claim as fraudulent, either due to filling errors or actual malintent. Our work was meant to be used as an exploratory tool to help inform/flag claims as fraudulent before fulfilling them. Fraudulent claims from human error is a very common issue in the industry.
Tools & Technologies
Python, SQL, scikit-learn