Establishing Data Science Thought Leaders

The Northwestern Mutual Data Science Institute is committed to solving important business and community challenges by harnessing the power of data for research and scientific discovery.

Our research efforts are prioritized by using data to address three major themes:

  • Tackling some of the world’s largest health crises
  • Addressing challenging social issues that affect a large population
  • Finding innovative solutions for business and industry problems

Underlying these areas is an unwavering focus on the ethical use of data to ensure we’re on the cutting edge of data science trends and effecting positive change in a responsible manner.

NMDSI Active Research

NMDSI-supported research advances innovative ideas, while fostering collaborations that will lead to long-term, deep relationships among faculty — across disciplines — at both UW-Milwaukee and Marquette. Through these collaborations, we will maximize contributions to data science while producing internationally-recognized data science research and thought leadership. Check out these current active projects:

The Elecurator Project

The project tracks the major issues engaging both candidates and voters in the 2020 presidential election cycle. Using a technique known as social curation, the team is analyzing multiple sources of data to see what topics are important to voters and how the candidates are speaking about those same issues.

Read More

Meet our Researchers

  • Dr. Vytaras Brazauskas
  • Professor and Associate Chair for Actuarial Science
  • Department of Mathematical Sciences
  • UW-Milwaukee

Dr. Vytaras Brazauskas’ research interests encompass actuarial science including credibility, loss models and pricing; quantitative risk management including operational risk and risk measures; and robust statistics including model risk, outliers and trimming.

In a series of papers that Dr. Brazauskas  jointly wrote with  his first PhD student, Harald Dornheim, they constructed robust credibility models that can be used for pricing insurance contracts. Those models were then employed to identify outliers, those data points that are inconsistent with the assumed model, and then used to design  an outlier-recycling algorithm. Implementation of the models on bodily injury data, Medicare costs, and real estate prices yielded remarkably accurate out-of-sample predictions. In this line of research, Dr. Brazauskas is most proud of how they were able to effectively solve most aspects of a business problem, starting with problem formulation and methodological derivations, proceeding with computer simulations and real data applications, and finishing with practical  recommendations.

Dr. Brazauskas would like his research to influence his discipline by reminding people to pay close attention to the vulnerabilities when making data-driven, model-based decisions. Specifically, he cautions on the need to remember that statistical inference, predictions and business decisions depend on the quality of data and assumptions needed to construct the model. He finished by saying that “all models are wrong but some are useful,” the quote often attributed to the statistician George Box.

When asked who he’d most like to collaborate with, Dr. Brazauskas named Arpad Elo, a former Professor of Physics at Marquette and a master-level chess player, with eight Wisconsin State Champion titles. In 1960, Elo developed a system for rating chess players, which is now known as the  Elo rating. Sixty years later, the system is still used in chess and has been implemented in many other areas including football, basketball, baseball, several board games and in various multiplayer e-sport video games. According to Dr. Brazauskas, the success of this system lies in its computational simplicity and intuitive formulas, which also helped to promote a perception that the ratings are fair.  Given Dr. Brazauskas’ research expertise and interest in chess and other sports, he would be excited to collaborate with Elo using today’s analytic methods and computing resources.

learn more

Affiliated Faculty

The NMDSI Affiliated Faculty program provides resources and support to data science faculty, tracks data science research and education, and facilitates building expert multidisciplinary teams from faculty with overlapping or complementary skills and interests.

By building a data science community based on collaboration and increasing the collective impact of data science research, the NMDSI Affiliated Faculty program brings us one step closer to achieving our goal of transforming our world through the power of data science.

Interested in joining the NMDSI Affiliated Faculty program? Submit your application below.

Apply Now

View Affiliated Faculty (PDF)