Review (approx time 20 mins)
We will start with a review of the first day:
Questions/comments on Insurance Analytics?
Questions/comments on R?
Questions/comments on Frequency Distributions?
The plan is essentially:
For each chapter, I will provide roughly a 45 minute introduction to the topic.
Then, you will work through the R code and data analysis for about 45 minutes.
We will then spend 15 minutes or so discussing the material as a group.
Modeling Loss Severity (approx time - 2 hrs)
Distributions that have intermediate and long tails are widely used in modeling insurance loss data sets. Moreover, these distributions, in particular the Weibull and gamma, are widely applied representations of survival times.
This segment reinforces the introduction that you received in your Probability and Statistics 1 course.
See: Modeling Loss Severity, Chapter 3 of Loss Data Analytics. See also the companion site, Loss Data Analytics with R
Nonparametric Inference and Model Selection (approx time - 1.25 hrs)
We will discuss the principles of nonparametric estimation and model selection.
See: Model Selection and Estimation, Sections 4.1 and 4.2 of Loss Data Analytics. See also the companion site, Loss Data Analytics with R
Estimation using Modified Data (approx time - 1.5 hrs)
We will discuss estimation for censored data, in particular, the Kaplan-Meier survival estimator. This estimator is highly utilized in the analysis of survival data.
See: Model Selection and Estimation, Section 4.3 of Loss Data Analytics (omit Section 4.3). See also the companion site, Loss Data Analytics with R
Pricing Insurance (approx time - 1 hrs)
Why bother with frequency and severity models? Setting insurance prices provide a practical motivation that is easy to interpret.
This segment reinforces the introduction that you received in your Econometrics: Linear Models course.
You are also likely seeing many of these ideas in your Underwriting and Reserves - IARD Insurance course.
See: Premium Foundations, Chapter 7 of Loss Data Analytics.