Data Science and Machine Learning Models Focus of MassMutual TECH Talk
On April 25, the Computer Science department, in partnership with MassMutual, hosted a TECH Talk that saw attendees learn how MassMutual’s advanced analytics team builds and deploys accurate and reliable models, using a mortality model as an example.
Guest speaker Sara Saperstein, who serves as lead data scientist in the risk and product domain of data science at MassMutual, led a discussion about the company’s data science team and how they leverage science and technology to build systems and create knowledge that enables data-driven decision-making across the entire company. In one example, machine learning models were used to improve the underwriting process for increased accuracy and a better customer experience. It’s critical that data science models in production are robust and performant, it was explained, so the advanced analytics team has developed approaches to ensure reliable service for the company’s end users and customers.
Ms. Saperstein was no stranger to Boston University, having been a PhD candidate in Cognitive & Neural Systems and Computational Neuroscience here at the University before leaving the program with a master’s degree to work as a data scientist at an ad tech startup, Pixability. She joined MassMutual in 2017, and in her current role she leads the effort to model mortality for the purposes of life underwriting, working closely with medical underwriting and data engineering teams to ingest relevant data sources, build more medically causally relevant models, and deploy them to stable environments for an excellent customer experience.