Invited Paper Session Abstracts - Mathematics for Data Science

With the abundance of data, new skills will be needed to prepare students for jobs that do not yet exist. Via the convergence of mathematics with other disciplines, such as engineering, we can enhance the student's ability to solve complex societal problems. This presentation will share recent advancements in the innovation ecosystem that utilize data science which are grounded in mathematics.

Data is a huge and growing part of our world. A major tool to manipulate and study this data is linear algebra. This talk will discuss concepts of matrix algebra that lay at a foundation of data science. We will see how to use linear algebra in such areas as compression, clustering, ranking, and text analysis.

This project was done as part of the MAA PIC Math program. PIC Math aims to increase awareness among math faculty and students about non-academic career options; teach faculty how to make industry connections to provide research experiences for their students to work on real problems coming directly from business, industry, or government; and prepare students for industrial careers. Over spring of 2019, I had three industrial collaborators from IntelyCare, Mitre, and AIR-WorldWide. IntelyCare is a nursing agency that aims to solve the per diem nurse-staffing issue by matching qualified nursing professionals with health care facilities. They have built an on-demand, tech-enabled, health care staffing platform that serve longterm facilities in six different states. The problem is twofold, first how to forecast the demand for nurses so that hospitals can have a better way to know how many substitute nurses may be needed for any time period. The second problem is to determine how long does it take for a nursing shift to fill at a given price. Several statistical analysis like machine learning and survival analysis are implemented to provide health care facilities with an understanding as to how they can increase the likelihood of their posted shifts being filled, and how long they can expect to wait for the shifts to be filled.

After eight years in academia, I transitioned into a data scientist role in industry. In this talk, I will draw on my experience in academia, my involvement as a faculty member in the PIC Math program, my work in industry, and a variety of conversations and short interviews with other data scientists in order to engage with the question of how mathematics plays a role in the field of data science. Topics of discussion will include foundational math classes, important mathematical techniques, and the data scientist skillset.

Nearly every question in data science is, or can be formulated as, an optimization problem. How can I carry out a task with the fewest resources? How can I minimize my risks? How can I do the most good? All of these are optimization problems. In this talk, I will address why a strong knowledge of optimization techniques is important, some of the key algorithms to know for data science, and best practices for teaching optimization to an aspiring data scientist.

Mathematical foundations is one of the key competencies of the 2016 Park City Math Institute’s Curriculum Guidelines for Undergraduate Programs in Data Science. A recent five-year program review of the Data Science program at Winona State University highlighted the necessity of additional learning outcomes centered on mathematical foundations. As a result, a new course is currently being developed to address the mathematical shortcomings of our existing curriculum. The topics to be covered in this new course will be centered around optimization, linear algebra, and probability. Methods and applications will be the primary focus of this course.