Vishesh Karwa, assistant professor of statistical science, joined the Fox School in 2018. Prior to his time at the Fox School, he served in the Department of Statistics at Ohio State University. He previously served two years as a postdoctoral fellow at Harvard University’s Center for Research and Computation for Society and one year at Carnegie Mellon University.
His research interests include data privacy; causal inference under network interference; social network models; algebraic statistics and computational methods for intractable likelihoods, among others.
How have you innovated in the classroom in the last year?
I’ve been working to make my courses more like workshops. I have deviated away from teaching with tools like PowerPoint and I’ve been using a software called Jupyter Notebook. Jupyter Notebook is an open-source web application that allows data scientists, and in this case, students, to create and share documents that integrate live code, equations, visualizations and other resources, along with explanatory text in a single document. We try to write a lot of code in class so that students can get hands-on coding experience. I’m always looking to make courses more interactive.
Most of my grading is based on projects instead of exams. I think the best way to learn and test coding skills is through projects, not straightforward examinations. So I have students collect real-world data sets online. Many of my students are using data from their own workplaces—they’re using the skills they are learning in class to solve problems they are experiencing at work. That made me really happy, actually. I’m glad that what I am teaching is actually useful for students right now in their current positions.
What are some of your current research projects?
I think of my research as a three-circle Venn diagram. The first circle I work in is data privacy, the second circle is network models and the third is causal inference. I work and research in the intersection of these three broad areas. Right now, one project I am working on is about genomic data—so, basically: how do we protect the privacy of individuals who are sharing their genetic information? And at the same time, how do you discover, let’s say, genetic markers for breast cancer or any genetic disease? I’m looking at how to make sure all of that data stays private and protected, but at the same time, used to learn something that can be useful.
Another thing I’m researching is voting habits and social networks—voting behaviors are not only dictated by advertisements or campaigns, but by peer influence as well. We call this peer influence causal inference in networks. So I’m basically doing research to disentangle this effect.
How does the MS in Business Analytics (MSBA) program prepare students, regardless of career path?
Our program is structured in such a way that we are teaching students practical, actionable skills. If you’re teaching a machine learning course to a PhD student, their goals are completely different—they need to understand the theory and the mathematics behind everything. But typically, in MSBA classes, their focus is more applied. They want to get their hands dirty and work with and program real data sets.
So there is a fundamental difference in the way you need to approach teaching for these two kinds of students. Our program is very well designed and catered to students who are looking to learn hands-on machine learning experience, hands on data analysis and data cleaning experience.