Publication: Comparativa de los modelos GLM y GBM para la tarificación de una cartera de autos
Loading...
Identifiers
Publication date
2022-06-15
Defense date
Authors
Advisors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
El sector asegurador se encuentra en una constante evolución en todos los ámbitos gracias especialmente a los avances tecnológicos. En la rama de la tarificación, la búsqueda constante de una mejor modelización del riesgo ha generado la creación y uso de distintos algoritmos para capturar de la manera más eficiente y exacta el riesgo de los clientes.
En este estudio se lleva a cabo la comparativa entre el método clásico de modelización lineal generalizado frente a la nueva corriente de modelos basados en Machine Learning llamados Gradient Boosting.
Los resultados obtenidos muestran una mayor capacidad predictiva para el método más novedoso y una mayor granularidad para ajustar las primas de los asegurados. Ambos métodos son válidos de cara al estudio académico, y observando los resultados, todo parece indicar que es el momento de iniciar el cambio utilizando con mayor asiduidad los nuevos modelos de Machine Learning.
También se lleva a cabo un análisis de negocio comparando una metodología de mutualización frente a una segmentación en los precios con el fin de mantener la cartera sana económicamente.
The insurance sector is constantly evolving thanks especially to technological advances. In the pricing section, the constant search for better modelling has led to the creation and use of different algorithms to capture customers risk in the most efficient and accurate way. In this study, we compare the classic generalized linear modelling method against the new trend based on machine learning called Gradient Boosting models. The results obtained show a greater predictive capacity for the most innovative method and a greater granularity to adjust the premiums of the insurers. Both methods are valid for academic study, and observing these results, everything seems to indicate that now is the moment to initiate the change and use more frequently the Machine Learning models. We also carry out a comparative business analysis between a mutualization and a segmentation price methodology in order to maintain an economically healthy portfolio.
The insurance sector is constantly evolving thanks especially to technological advances. In the pricing section, the constant search for better modelling has led to the creation and use of different algorithms to capture customers risk in the most efficient and accurate way. In this study, we compare the classic generalized linear modelling method against the new trend based on machine learning called Gradient Boosting models. The results obtained show a greater predictive capacity for the most innovative method and a greater granularity to adjust the premiums of the insurers. Both methods are valid for academic study, and observing these results, everything seems to indicate that now is the moment to initiate the change and use more frequently the Machine Learning models. We also carry out a comparative business analysis between a mutualization and a segmentation price methodology in order to maintain an economically healthy portfolio.
Description
Keywords
Modelización, Machine Learning, Tarificación, Negocio