Basics of Non-Life Pricing in R (R5)

This course will allow you to explore utilising building a technical tariff with generalised linear models (GLMs) in R. 

The course starts with an introduction to risk classification, introducing portfolio heterogeneity, risk classification, technical vs commercial risk pricing, and the process. Thereafter it explains the need for regression and the move from linear regression to GLMs, introducing a variety of GLM families.

We then explore Poisson regression for claim counts as a frequency model, how to code categorical explanatory variables, how to interpret the likelihood equations, finding the variance and deviance, testing the hypothesis, and analysing the claim frequencies.

Lastly, we consider overfitting and how to assess the relevance of models, introducing the following as severity models:

• Gamma regression for attritional claims 
• Logistic regression for claims occurrence 
• Extreme value theory for large claims modeling
  

The course ends with the live lesson which deals with the practical difficulties of GLMs. There is reference to two practical experiences, the example and the hands-on case study. The example showcases Poisson, Gamma and logistic regression in R, whereas the case study is a hands-on experience in developing a new technical tariff. Both Notebooks, along with a memo, are available for completion on the platform.

This course is presented mainly through a combination of videos, slides, and Jupyter Notebooks. After each video, there is a short quiz for the student to gauge their understanding of the section before continuing.