Competition Master

Q&A with Carlos Brioso, FSA, CERA, director, Center for Data Science and Analytics, New York Life August/September 2019

Photograph: Rob Tannenbaum

You recently earned the title of Kaggle Competition Master. What keeps you coming back to Kaggle’s data science competitions?

There are several reasons why I keep coming back. First, this a great way to apply my analytical skills—I get to solve a variety of problems and be measured in a very objective way. Second, there is an active interaction with a very talented community that provides a feedback loop, enriching my experience and knowledge. Third, the competitions give me the opportunity to develop different ideas and apply state-of-the-art machine learning techniques.

How did you get involved in Kaggle competitions?

My previous employer organized an internal Kaggle competition. The objective was to estimate the expected insurance payments for claims in a business with long-tail risk. My team finished third in the competition. The proposed solutions considerably outperformed traditional actuarial methods used for reserving. This demonstrated the value of using new methodologies and software to tackle traditional problems in a different way.

What does it take to be successful?

Most people have the impression that applying new algorithms and having a powerful computer available are what make the difference. These things are important and help, but the key to being successful in these competitions is to do the basics of modeling correctly: Know your data, try to understand the problem, get some domain knowledge and learn from what others have tried. It is also important to understand what the advantages and disadvantages of the different algorithms are and use them tactically. Usually, a single algorithm is not sufficient. Combining different approaches to the problem and using a variety of algorithms are the keys to succeeding.

What can actuaries do to position themselves for data science roles outside of the insurance industry?

Actuaries are good at understanding business problems and using analytics to make business decisions. These skills are sought-after by data science teams. Actuaries should acquire additional skills such as programming and machine learning to be competitive outside of the insurance industry and participate in projects outside of the insurance domain. Credit risk is very close to insurance in terms of the types of problems that need to be tackled, so this is a good area to begin exploring problems outside of insurance.

What is your best advice for staying ahead of the competition?

Look for areas where you can provide value in the industry. Update your skill set. Step outside of your comfort zone, and you might be surprised how you can excel in new areas.

What is the most satisfying facet of your current position at New York Life?

Our data science group is trying to address different business problems with new techniques and data. The satisfaction comes from knowing our work can provide value to the company and our customers.

Describe a successful day.

A successful day is when we strike a good balance between the complexity of our technical work and the effectiveness in communicating these results to our business partners.

Carlos Brioso, FSA, CERA, is director, Center for Data Science and Analytics, New York Life.