Predictive Analytics and New Opportunities

Photograph: Hyon Smith

Predictive analytics, big data and predictive modeling are terms most of us have undoubtedly heard, especially during this past year. The discussion of these topics has certainly evolved quickly from “what is it?” to best practices and new ideas for practical applications. As actuaries, we know data and modeling. We can argue that we are the original users of big data. We can provide a comprehensive approach to build and validate models within our industry and in other industries, too.

The Society of Actuaries’ (SOA’s) Cultivate Opportunities Team (COT)¹ has been a big part of our efforts to create a foothold for actuaries in this field, both inside and outside traditional actuarial employment. Also, those working on the Learning Strategy have developed a number of curriculum changes designed to better prepare actuaries to be successful in these fields.

I would like to share some of the advancements of practicing actuaries and also the SOA’s preparations for future actuaries. We must look at today’s challenges as well as tomorrow’s.

At the start of the year, the SOA launched an advertising campaign on predictive analytics to reach health employers. We developed a brochure² to emphasize the work of actuaries with health care predictive modeling and data analysis. This work helps employers and recruiters understand the roles actuaries can play within health care, especially when it comes to predictive analytics.

Did you know we have offered actuarial internships to employers outside of the insurance industry? Our internships help expose employers to the knowledge and skill sets actuaries can bring to their companies. For example, Validate Health shared that its intern’s theoretical knowledge in computer science and actuarial science was concretely applied to model development within a provider managed care context. We are glad to have this feedback, and we secured several more internships this past summer with Microsoft, NASA and a number of health companies that utilize data analytics.

In June 2016, the SOA Board of Directors approved changes to the ASA curriculum. One of the important drivers is predictive analytics. Under our new approach, candidates will use computer packages to learn—in a hands-on way—how to apply the principles of predictive analytics with realistic data sets.

For practicing actuaries, we offer continuing education concurrent sessions, webinars and other materials on predictive analytics. Soon we will offer a certificate pilot program in self-study and assessment for fellows who want to make a material commitment to learning this material. The pilot will be open to a limited number of participants, and it is geared toward actuaries with five or more years of SOA membership. We currently are building this program as a relatively deep dive for actuaries in building their skills with predictive analytics. Stay tuned for more information in 2017.

Our organization continues to develop and fund new research and thought leadership essays on predictive analytics and related work. The SOA recently released a new collection of papers³ covering:

• Adoption of machine learning techniques, which is a field of predictive analytics focused on ways to automatically learn from the data and improve with experience.
• A practical example of the intersection of insurance and machine learning with mortality prediction.
• How to compare policyholder efficiency in variable annuity lapses.
• The artificial neural network model and how to estimate the probability of new insurance purchases.

On LinkedIn,⁴ I have been sharing articles discussing interesting trends and developments involving our profession, including the applications of predictive analytics. I encourage you to take a look at some of those predictive analytics conversations, such as on climate change, the SOA 2017–2021 Strategic Plan and the curriculum changes for the associate pathway. The webpage SOA.org/predictive provides resources, case studies and new ideas from our members working with predictive analytics.

Please share your ideas with me, as I appreciate hearing your perspective.