Applications of Clustering with Mixed Type Data in Life Insurance

Death benefits are generally the largest cash flow items that affect financial statements of life insurers; some may still not have a systematic process to track and monitor death claims. In this article, we explore data clustering examine understand how actual claims differ from what is expected—an early stage developing monitoring system crucial for risk management. We extended k-prototype algorithm draw inferences insurance dataset using only insured’s characteristics policy information without regard known mortality. This has feature efficiently handling categorical, numerical, spatial attributes. Using gap statistics, optimal clusters obtained then used compare expected experience portfolio. Our empirical contained observations approximately 1.14 million policies with total insured amount over 650 billion dollars. For portfolio, produced three natural clusters, each cluster having lower but differing variability. The analytical results provide management identify policyholders’ attributes dominate significant mortality deviations, thereby enhance decision making taking necessary actions.