Approaches Towards A Recommendation Engine For Life Insurance Products

Abstract

Recommender engines are powerful tools in today’s world to overcome the problem of over choice. As the world is moving towards information overload, the life insurance industry is no more immune than any other domain. Three broad categories of life insurance plans are namely – Endowment, Term and ULIP. This paper discusses a variety of ML models that aim to classify the right fit product category for a new customer (extendable to existing customers) on a real-time life insurance company dataset. The dataset used for the modelling were of 2 kinds. The first kind contained features of customer demographics – age, location, education and occupation. The second dataset included these customer demographics as well as the bureau information of the respective customers which included multiple features describing their credit history. By the means of clustering, collaborative filtering approaches were tried on. Also, the problem was tackled using predictive modelling techniques such as Random Forest, Decision Trees and XGBoost.