Exploring Functional Patterns of Driving Records by Interacting with Major Classes and Territory Using Generalized Additive Models

Shengkun Xie, Anna T. Lawniczak, Clare Chua-Chow

Abstract: Studying the safe driver index, such as Driving Records (DR), is essential to auto insurance regulation. Part of the auto insurance regulation aims to estimate the relativity of major risk factors, including DR, to provide some benchmark values for auto insurance companies. The risk relativity estimate of DR is often through either an assessment via empirical loss cost or a statistical modelling approach such as using generalized linear models. However, these methods are only able to give an estimate on an integer level of DR. This work proposes a novel approach to estimating the risk relativity of DR via generalized additive models (GAM). This method makes the integer level of DR continuous, making it more flexible and practical. Extending the generalized linear model to GAM is critical as investigating this new method could enhance applications of advanced statistical methods to the actuarial practice. Thus, making the proposed methodology of analyzing the safe driver index more st

External IDs:dblp:conf/data/XieLC23