Introduction to Generalized Linear Models for Insurance Pricing
Details
By: Sam Kloese
Insurance companies do their best to charge premiums that are proportional to the risk of the insured. Generalized Linear Models (GLMs) are a common method of helping determine the relative level of risk for different classes of business. GLMs have the advantages of being easily explainable and programmable, come with several statistical tests to justify their significance, and have a lot of precedence for their use in insurance.
This session we will give some background on GLMs and then dive right into building GLMs with R. I will demonstrate how GLMs are an obvious improvement over univariate methods using a theoretical example. Then I will give a walkthrough of GLM modeling decisions on a real dataset. This presentation will rely on the following packages: tidyverse, ggplot2, and insuranceData. The real dataset case will be based on the dataOhlsson table from the insuranceData package.