Predictive Analytics Case Study: Dorothy Andrews, ASA, MAAA
Since this interview, Dorothy Andrews has taken on a new role as chief behavioral data scientist at Insurance Strategies Consulting.
Dorothy Andrews has always sought out opportunities that require statistical and data analytical skills, especially in nontraditional spaces. Her philosophy has always been "learn the business you want to build predictive models for."
Today Dorothy is a consulting actuary with Merlinos & Associates, a leading property, casualty, and health actuarial consulting firm. While her skills translate to a variety of industries and models, she works with her life insurance clients to build predictive analytics marketing applications to better segment customers.
"Just the way Netflix suggests viewing picks to its subscribers based on past viewing selections, marketing insurance products is more effective when you customize the marketing of products to consumers most likely to buy them," she said.
Dorothy did not start her career to be an actuary, however. In fact, the former high school math teacher didn't learn about the profession until she returned to graduate school. She was fascinated with how to use statistics to solve real world problems, which pushed her to learn new skills.
During her tenure at John Hancock Mutual Life Insurance Company, Dorothy partnered with IT professionals to help them interpret and correctly program actuarial solutions. That is where she learned about IT architectures and protocols.
She was then one of the few actuaries working for Wachovia bank (now Wells Fargo) and the only actuary in the credit risk area. She worked closely with mathematicians, statisticians and economists in what she calls a "rich intellectual environment."
Yet it was in her role at the United States Department of Agriculture (USDA) that Dorothy truly fell in love with predictive analytics. She was part of the team building the predictive analytics engine for the USDA's new Public Health Information System. She took pride in helping to build a food safety inspection system to monitor and protect the nation's food supply from threats like residues and pathogens.
When Dorothy returned to insurance, it was to lead a team of predictive analysts for a property & casualty (P&C) insurer. Recognizing the synergies between the predictive modeling, underwriting and actuarial pricing areas, she worked with her team to replace outdated models with new, trusted models to understand key areas of the business.
Today it's easy to see the role of predictive analytics in P&C and auto insurance, especially given the large data sets from the frequency and variability of claims. Yet life insurance can appear too different to use the same techniques.
"Not to be too simplistic," Dorothy said, "but there is only one claim on a life insurance policy and that claim terminates the policy."
However, Dorothy sees predictive analytics as increasingly important in life insurance, especially as the industry tries to attract younger customers. Predicting policyholder behavior can enhance underwriting, marketing assumption settings and financial modeling. By coupling lifestyle attribute data and medial underwriting data, insurers can assess customers' healthy lifestyle choices, for example, and they can provide better, technology-driven products to the right customers.
Regardless of the industry, Dorothy believes modelers need strong theoretical skills and an inquisitive mindset to translate data into relatable solutions.
"A modeler needs to be able to distinguish between statistical noise and true statistical signal in underlying data," said Dorothy. "They also need developmental questioning skills and an inclusive communication style because models don't build themselves; people do."
Dorothy thinks actuaries are inherently qualified to lead predictive analytics efforts changing the industry. She says opportunities are everywhere for actuaries if they are not afraid to venture beyond traditional boundaries.