Analysis of Prudential Life Insurance Customer Evaluator

Abstract

Abstract: Life Insurance is one the most important part in today’s world. And with many varieties of insurance policies, the objective is to provide the most suitable policy which is beneficial to both the customer and the insurance company. This is generally done by the insurance agents based on their domain knowledge. But our goal here is to do it using a statistical solution. One method is to use Customer Scoring based on information provided by them when filling their insurance application form. Based on this score they are provided a set of insurance policies to select from. For this we tried building regression models (Linear and Multinomial Logistic) based on the customer information and scores we already have. These scores were provided based on the policy they already have. Different models based on different variable combinations were compared using Stepwise AIC Method for both regression models. The final model has an accuracy of 44% which can of course further be improved. This kind of statistical modelling will be useful in filtering the large number of policies to select from. After which the customer or agent may select from the smaller number of choices suitable for them. This will make the job of the agents as well as the customer much easier. Keywords: Life Insurance, Customer Scoring, Statistical Solution, Logistic Regression, Multinomial Logistic Regression, Stepwise AIC Method