With roughly one-in-three insurance companies expanding or piloting data science units in 2016, they may be facing an unanticipated problem—friction between their actuaries and data scientists, according to a research firm.
In a report announced last month titled, “Data Science and Actuarial: Managing Potential Conflict,” Novarica, a research and technology strategy consultant, reviews the reasons for conflict and suggests ways of getting actuaries and data scientists to work synergistically to provide greater value for their companies.
“With both functions deriving their organizational value from the ability to turn data into insight, there is a high potential for conflict between actuaries and data scientists,” said Mitch Wein, Vice President of Research and Consulting at Novarica and lead author of the report, suggesting that data scientists may view actuaries as “traditionalists” and actuaries may see the data scientists as “dangerous cowboys.”
While Novarica says that data science is simply a new term for an old thing—analytics and statistics—actuaries and data scientists approach their analytical and statistical tasks with different tools and training. Actuaries are trained in math (statistics and probability and calculus, mainly). They use spreadsheets (Excel, for example) and some programming skills (with knowledge of SAS or VBA or C++, in some cases).
Data scientists approach growing data sets of “a billion or more points” with tools that Excel spreadsheets can’t handle, bringing significant coding experience to bear and using programming languages such as Python and R.
The report details some signs of conflict within an organization—lack of clarity about leadership of data projects, lack of clarity about ownership of structured and unstructured data, and unclear expectations overall. Commenting on the latter, Novarica suggests that insurer executives are one source of problems in that the executives think that “data science is more flash than substance.”
“Without buy-in, the support necessary for a flourishing data science program will be very difficult. Leaders must clearly set expectations for where data science will be used, and what types of potential value they expect to generate,” the report says.
While a key finding of the report is that the conflict is common, it’s also manageable, Novarica says, offering three models for data science units.
- Setting up a standalone data science unit.
- Forcing actuaries and data scientists to work together.
- Embedding data scientists within the actuarial function.
Novarica comments that the last alternative may take longer to work, and outlines the benefits and drawbacks of the other options as well.