Registration is now closed. The SOA Customer Service Center is available to assist you Monday through Friday, 8:00 a.m. to 5:00 p.m. CDT. Please call +1-888-697-3900.
SOA Predictive Analytics & Futurism Section
In this hands-on one-day seminar, participants will learn how to build a basic predictive model through generalized linear models using R. They will learn how to build the models for life and annuity experience analysis; how to apply the models in assumption development for pricing and valuation; and their differences from traditional models.
Experts will introduce the key concepts of predictive analytics modeling like translating a question into a model form, finding the right balance of model complexity and interpretability while avoiding overfit. Participants will gain an appreciation of some of the mathematical concepts underlying the modeler’s choices. They will learn how to communicate their modeling results to a less technical audience in a clear and understandable way. Finally, participants will be exposed to more advanced techniques, as well as the tradeoffs among the range of platforms and languages available.
The R software package will be used for this seminar. Participants are expected to go through a brief introduction to R prior to the seminar. Software and dataset download locations and background material will be distributed in advance.
In order to participate fully during the seminar, please be sure to:
- Bring a laptop with a 64-bit OS
- Install required software in advance, including R and RStudio
- A complete list will be provided well in advance
- A tech session the night before the seminar will be provided for troubleshooting
- Download the sample datasets
- Complete basic R training sufficient to run R and execute the demonstrations
At the end of the seminar, participants will be able to create a basic predictive model using the R software package, evaluate it and communicate the results to a less technical audience. Participants will also have an understanding of the resources available to them for extended applications of predictive modeling, including software platforms, other popular languages and options for automating documentation.
This seminar is ideal for anyone who is interested in learning about predictive analytics and would like to have a deeper understanding of the subject by learning with a hands-on approach.
Level of Difficulty
The seminar is designed for participants with beginner level experience in predictive modeling; current familiarity with ASA-level statistical concepts such as likelihood maximization and minimization of mean squared error; and ability to pick up R syntax.