Model for the Detection of Potential Car Insurance Customers by Applying Machine Learning Algorithms

Abstract

The reduction in the production rate of the insurance sector in Peru is a problem that not only affects insurance companies, but also citizens, who cannot protect their valuable assets. This is due to the low acceptance of insurance in the Peruvian market and the inefficient use of the large amount of information and current technologies to detect clients by companies in this sector. That is why the need arises for a predictive model to identify which people are more likely to purchase insurance and what variables influence it. To do this, in this study, a solid model based on Machine Learning is proposed using data from a Peruvian insurance company, which is made up of sociodemographic variables and the vehicle that the person owns. The results show that the variables that influence being a potential insurance client are “employment status”, “educational level”, “salary range” and “vehicle size”. Also, it was identified that the Decision Tree algorithm obtained the best performance, with an Area under the curve (AUC) of 0.88 and an F1-score of 0.82 in predicting potential clients.