A Comparative Analysis of Fraud Detection in Healthcare using Data Balancing & Machine Learning Techniques

Frauds in healthcare sector has become a major problem and is widespread now. It is reported that total healthcare expenditure is increasing exponentially due to fraudulent claims in health insurance sector. It is a planned crime that include providers, doctors/physicians and beneficiaries etc, who together perform certain actions to make fraud claims. Most vulnerable institutions due to such frauds are the insurance companies. In this work as large data set is used so we have performed Exploratory Data Analysis (EDA) and then data preprocessing & feature engineering to create a feasible dataset for further analysis. In this proposed methodology, we have depicted a comparative analysis of results of different machine learning models using two data balancing techniques, i.e Class Weighing Scheme (CWS) and Adaptive Synthetic Oversampling (ADASYN) for oversampling along with the results of these models with unbalanced data set. Our objective is to verify the efficacy of various machine learning models in this domain.