Q&A with Robert Eaton, FSA, MAAA and Consulting Actuary at Milliman
As an actuary working in predictive analytics, how do you see predictive analytics changing in the next 5-10 years?
I see the role of predictive analytics expanding greatly. It should definitely become a skill that any actuary has in their toolbox. At our firm, we use predictive analytics in life insurance underwriting to set assumptions about policyholder behavior. We also use predictive analytics in long-term care. In the future, the use of predictive analytics will expand to apply to sales and marketing in the life insurance industry, and also to the claims department for fraud, waste and abuse discovery.
How has the SOA predictive analytics certificate program benefited your team and your clients?
I'm able to better manage projects with my team that involve predictive analytics and have shared key learnings from the program with my colleagues. For our clients, our team now has a broader base of actuaries who are able to use predictive analytics in their day-to-day work, and we can apply predictive analytics to more of the problems that are facing our clients.
How do you think the predictive analytics certificate program can improve a company's actuarial department and organization overall?
First, there's a certain structure to predictive analytics and how you build models. This is a foundational element of the certificate program, and during the program we spent a significant amount of time building models from the ground up. Actuaries can benefit from that structure and learn how to pull in data in order to generate results.
Second, the results that come out of predictive modeling can give actuaries new data to work with that will yield better results. For the past several decades, actuaries have used traditional experience analysis approaches and have viewed results and assumptions on that basis. During the program, we experienced first-hand that using predictive analytics gives actuaries more data to consider, which in turn may produce better outcomes.
Finally, a focus of the certificate program is that predictive analytics can help in other areas of the business as well. For example, predictive analytics can help underwriters make better decisions, and the claims team may be able to more easily detect fraud, waste and abuse.
Having been through this program, I feel strongly that predictive analytics is the future for actuaries and something we as an industry need to embrace.
What would you say to actuaries considering enrolling in the predictive analytics certificate program?
The program is a challenge and it’s very hands-on. It’s important to understand that, and if you're enrolling in this course, you will think differently afterwards. You will approach your work as an actuary using predictive analytics in a different way after completing the program.
How has the program enhanced your knowledge of predictive analytics?
The predictive analytics certificate program gives a very strong theoretical background and generally enhanced my knowledge of how to apply predictive analytics to my work. I learned a lot of underlying theory in the program to demonstrate why these predictive models work and what they're teaching us, as well as how to execute on that and actually build real models to apply to my current role.
How have you applied what you've learned from the program to your work?
The skills I learned in the predictive analytics certificate program are directly applicable to the work that I do each day. As an example, our team often looks at long-term care claim duration. When people go on a long-term care claim, we estimate how long they stay in a nursing home or facility, or in home care. Actuaries traditionally analyze that information by the age at the time they go on claim, their sex and where they began their claim. Using predictive analytics, we are able to take that data, merge it with other data, and take a hard look at additional drivers of how long people stay on claim. This use of data and modeling leads us to question if age and sex are really the best drivers of understanding that assumption.
What we found is those factors are pretty good indicators, but there may be are other considerations we should be looking at. For example, geographic area may be a proxy for income, and may have an impact on the claim duration. Predictive analytics has helped us understand that there might be more factors out there that we should consider.
What do you think is different about the predictive analytics certificate program compared to the other curricula offered by the SOA?
When I took my actuarial exams in the mid-2000s, we had a series of about six exams and then a set of modules to get your ASA and FSA. The predictive analytics certificate program most resembles some of those modules. It's very hands-on, and it causes you to produce results yourself rather than learning something without the real-world application. While it’s incredibly important to lay the foundation of your actuarial knowledge with those exams and modules, the predictive analytics certificate program is a rigorous next step to advancing your skills.
What other benefits did you get from the program?
After completing the program, I can now speak the predictive analytics language. There are a few great predictive analytics thinkers in the actuarial industry, and when they publish papers, I can understand the material. What’s more, I can see how some of what they write might apply to my own work.