With the excitement around AI and algorithms it’s easy to lose sight of the human intelligence that’s driving transformative property imagery and analytics. Madeleine Lopeman, Betterview’s Lead Data Scientist, exemplifies the talent and education the insurance industry needs to contribute to risk management in light of climate change.
Madeleine received her PhD from Columbia University in Civil Engineering and Engineering Mechanics. Her doctoral dissertation focused on quantifying and assessing storm surge risk in Lower Manhattan. After graduating from Columbia, she joined reinsurance broker Guy Carpenter as a part of parent company Marsh & McLennan Companies’ Innovation Center. When that project ended, she joined Blackboard Insurance as a Data Scientist. I recently talked with her about data science, her transition from civil engineering to data science and insurance, and her leadership role at Betterview.
First, what is a data scientist? How and why did you become one?
I studied civil engineering at Columbia University and came into data science and statistical modeling while working with my PhD advisor George Deodatis, Chair of the Department of Civil Engineering and Engineering Mechanics. For my doctoral research, I built statistical models using tidal gauge data from downtown Manhattan to determine whether Hurricane Sandy’s flooding was a once in lifetime event for New York City or if it could happen more frequently than the state-of-the-art models at the time indicated.
How did you get from civil engineering to insurance?
Through my research at Columbia, we found that the risk of storm surge in New York City may have been underestimated prior to Hurricane Sandy and that the risk would intensify as sea levels continue to rise. Armed with this refreshed and more comprehensive view of the potential risk of coastal flooding to New York City’s infrastructure, we began thinking about ways buildings, subways and tunnels could be adapted to withstand the impact of storm surge on a low-lying and dense city like New York. It was clear throughout the process that our work had important implications for planning, insurance and engineering decision-makers.
After completing my doctorate, I joined Guy Carpenter to become part of a team at the company’s innovation hub in Dublin, Ireland. Our mission was to build tools to evaluate the major catastrophe risk models used in the insurance industry by comparing them to independent scientific research. We collected the scientific and insurance data into a unified framework and built applications to generate data-driven insights to inform (re)insurance portfolio risk management and decision-making.
In 2019, I joined Blackboard Insurance as a Data Scientist where I built predictive models using third-party data to analyze and understand insurance risks.
Why did you decide to join Betterview? Did you know anything about the company prior to joining?
I became acquainted with Betterview through my time at Blackboard. Right away I was impressed both with Betterview’s product offering as well as its way of working, especially its focus on understanding the client’s needs. Through using Betterview’s remote property intelligence platform and exploring their data, it was clear to me that Betterview was tackling data and machine learning issues facing the insurance industry in a smart and targeted way. Moreover, Betterview’s transparency about its data and models facilitated trust and allowed us to rely on its data to support insurance decisions.
For all of these reasons I was thrilled to have the opportunity to join Betterview and develop data science products to advance modern property risk management. Since joining Jason Janofsky’s engineering team, I’ve also come to appreciate Betterview for the opportunities it has afforded me both to contribute to and learn from a well-oiled data science machine and to partner with insurance carriers to solve problems together.
How do you think your career experiences at Guy Carpenter, Blackboard Insurance and your civil engineering background ties into the work you do at Betterview?
At Betterview I’m responsible for data collection and model evaluation, two threads that have woven throughout my career. Through a close partnership with Betterview’s Lead Computer Vision Engineer, Julius Simonelli, we build models to inform property risk management and decision-making.
In order to build a model to detect a certain roof condition, we need to collect a large number of images of that condition, which may come from a variety of sources. Each condition presents a new research challenge that requires drawing on my experience analyzing third-party data sources as well as research skills developed through my doctoral program. For example, we’ve developed algorithms to scour aerial imagery immediately following hurricane landfalls to collect thousands of examples of structural damage.
Once we’ve collected this data, we hand it to Julius’s team, who builds the computer vision models that analyze visual patterns on roof tops to detect risk indicators. After the models have been developed, Julius passes them back to the data science team for an independent and unbiased evaluation. This “model arbiter” function is a role that has carried through various elements of my experience in risk assessment.
Through our analysis and partnership with the computer vision team, we set the bar for model quality and ensure continuous improvement of Betterview’s visual detections and scores.
How do you see Betterview helping insurers and clients with climate change?
There are two classical strategies to develop climate change resiliency: adaptation and mitigation. Climate change mitigation involves addressing the causes of the risk and minimizing their effects (e.g., reduction of carbon emissions). Climate change adaptation is focused on reducing the impact that the changing climate can have on the built environment and society (e.g., flood walls). Betterview can directly assist clients – insurers and property owners – in proactive adaptation efforts by showing them properties and helping them proactively detect potential risk indicators prior to a loss.
What do you do outside of work? What would people be surprised to learn about you?
My husband and I met while training for the New York City marathon. We’re not serious athletes but we’ve run a marathon every year for 6 years, though unfortunately this year we had to put that on hold because of the pandemic.
As for what would be surprising to learn about me – I come from a family of artists and musicians. I’m the only engineer in the family.
What’s the best way to contact you?
You can reach me via email at email@example.com or connect with me on LinkedIn.