Skip to content

RonRichman/pc-pricing-tutorial

 
 

Repository files navigation

Travis build status Gitter chat

P&C Pricing Tutorial

The goal of this project is to build an end-to-end reproducible example of a ratemaking project in R, in the form of a series of blog posts. The target audience includes students, actuaries, and data scientists who are interested in learning about insurance pricing or porting their existing workflows.

A live preview of the tutorial can be found at https://pricing-tutorial.netlify.com/.

Scope

While we won't be able to emulate all the beauracracy/politics associated with doing data science projects in an insurance company (they'll be different for everyone anyway), we'll comment on how they might affect the analytics workflow. As much as possible, we'll provide reproducible code for the technical bits, including data manipulation, exploratory data analysis, modeling, validation, implementation, and report writing. Significant simplifications from real life (due to lack of details in the dataset, for example) will be noted. We'll follow modeling best practices, but also point out incorrect/suboptimal workflows that are prevalent. Recognizing that the output of a pricing project will undergo regulatory review, we'll also monitor the development of the NAIC draft on Regulatory Review of Predictive Models and adapt accordingly.

Data Access

We're using the publicly available data from the Superintendência de Seguros Privados of Brazil. The data can be downloaded from http://www2.susep.gov.br/menuestatistica/Autoseg/principal.aspx. We'll be using the first half of 2012 (Autoseg2012B) to start, since that's the first time period where CSVs (as opposed to MS Access files) are available. The convention we'll use in this repo is that we'll put the uncompressed directory in external_data/, so your directory structure might look something like

├── R
├── analysis
├── external_data
│   ├── Autoseg2012B
│   └── gadm36_BRA_gpkg
├── man
└── manuscript
    ├── images
    └── public

The .gpkg files for drawing state boundaries can be downloaded from https://biogeo.ucdavis.edu/data/gadm3.6/gpkg/gadm36_BRA_gpkg.zip. See https://gadm.org/download_country_v3.html for details.

Contributing

Interested in joining in on the fun? Look at the issues page to see what tasks need help and check out contributing guidelines. Not familiar with R but want to lend your actuarial expertise? Please feel free to comment on issues to share your thoughts or open new issues to let us know how we can do things better!

Releases

No releases published

Packages

No packages published

Languages

  • HTML 84.6%
  • JavaScript 14.0%
  • Other 1.4%