Bridging the Gap: The Capstone Experience

Hello! My name is Anthony Battista, and I am an MBAn student in the Class of 2019! Just last year, I graduated from UMass Amherst, where I earned a bachelor’s degree in civil engineering with a minor in mathematics. During my undergraduate years, my chief interests centered around transportation engineering, statistics, and computer science. I came from a particularly academic background; I was heavily involved in research, and I worked in multiple research groups that introduced me to the analytical aspects of transportation.

As I approached my final year of undergrad, I began to realize that my strongest passions were for operations research and analytics. I decided to pursue graduate education in these fields so that I could delve in deeper and further develop my skills. Initially, I had planned to go directly into a PhD program upon graduation; after much consideration, though, I ultimately decided that I wanted to start by pursuing a master’s degree and gaining some industry experience first. One thing led to another, and now here I am at my lifelong dream school, MIT!

One of the main things that drew me to the MBAn program was its professional focus. Given that my pre-MBAn work experience had been predominantly academic, I really liked the fact that MBAn presents so many opportunities to work on real projects and make connections in industry. My cohort got a taste of this last semester with our Analytics Lab class, where we all partnered up with companies and spent the semester working on data science projects for them. The centerpiece of MBAn, though, is the Capstone Project – where we partner with our classmates into teams of two, get matched up with companies, and intern for them as data scientists from January through August. We work on our projects part time throughout the spring semester, and then move on-site to intern full time once the summer begins.

My partner Noah and I are currently interning as data scientists for reinsurance company Swiss Re. Going into the capstone matching process, neither of us possessed any prior experience or familiarity with reinsurance – let alone with the insurance industry in general! In fact, we were initially not quite sure what to expect from an insurance data science project at all. When we saw Swiss Re’s presentation on Capstone Pitch Day, though, our interests were piqued. After interviewing with them and learning more, we became very interested in their project, which was pleasantly open-ended and presented numerous avenues for exploration. Fortunately, we ended up matching with Swiss Re, and have been working with them since then.

Reinsurance companies often tend to tackle some very interesting broad problems – for example, last year’s Swiss Re Capstone team utilized machine learning to predict wildfire hazards in North America. Swiss Re’s project this year focuses on addressing the life insurance protection gap – that is, the gap between the amount of life insurance families should have to protect their income and the amount of life insurance they actually have.

The company’s ultimate mission is to make the world a more resilient place, and a major step towards accomplishing this is addressing the protection gap. In particular, we are interested in understanding what sorts of factors inspire people to purchase life insurance, whether or not these factors vary over geography and time, and how this knowledge can be used to improve insurance marketing.

So far, we have had the chance to implement a variety of techniques, including predictive modeling, cluster analysis, time series analysis, and dimensionality reduction. In the near future, we even plan to incorporate techniques such as natural language processing with public social media data. It’s been very exciting to see these techniques from our classes “come to life,” so to speak, on an actual project that we hope will have lasting positive impact on Swiss Re.

Given my prior background, Capstone has been an invaluable experience for me. I’ve quickly learned a lot about the workings of the business world, and it’s quite a different place than the academic world that I recently inhabited. For example, while much of my past work focused on developing and improving theories, now my project goals are centered around producing actionable, tangible business impact. I’ve gained firsthand experience navigating varying needs and requirements posed by different industry stakeholders, and I have now come to understand the challenges and victories that come along with working on a diverse data science team.

Above all, I have had the chance to work closely with “real world” data, which is vastly different than the carefully-curated, frequently-pristine datasets that tend to be used in data science instruction. Academic datasets are generally error-free and are often carefully engineered to hold some predefined “message.” This is rarely the case with real data. Real data is messy. Sometimes it has been entered by humans and shows clear signs of human error; other times it may be missing large amounts of important information, necessitating careful and creative handling approaches. All-too-frequently, the key message of the data is heavily obfuscated – or may even be nonexistent. For all of these reasons, working with real data has been an exciting challenge that I’ve learned a lot from.

Swiss Re’s main American office is located in Armonk, New York. Over the course of the spring semester, we have been working on our project remotely from Cambridge under the guidance of our faculty mentor Dr. Carine Simon and our PhD student mentor Hari Bandi. Today, we began our work on-site in their office. I am very excited to get to know the rest of the data science team and to spend the summer living in New York City! I’ve had a great experience working for Swiss Re so far, and I look forward to the weeks ahead.

Leave a Comment

Your email address will not be published. Required fields are marked *