At the Argyros School of Business and Economics, Dr. Jonathan Hersh, Assistant Professor, Economics and Management Science, teaches courses in machine learning, statistics and business analytics with an emphasis on applications for business. Dr. Hersh received his Ph.D. from Boston University. His research focuses on the economics of information systems, development economics, data-driven decision making and applied artificial intelligence.

Before Chapman University, Dr. Hersh was a lecturer at both MIT and Wellesley College. He has also consulted with the World Bank, Inter-American Development Bank, and the International Monetary Fund and was able to demonstrate how management processes could be improved with the application of machine learning.

Q&A with Dr. Hersh

What research or publication are you most proud of completing? Why?

I think the publication I’m most proud of completing is Poverty from space: using high-resolution satellite imagery for estimating economic well-being. In this paper, we applied machine learning to satellite imagery to extract satellite features, which were then used to model poverty.

This type of research was very speculative and novel at the time; many people said that it was unlikely to work or would be infeasible. Whenever you do something that is that new and different, there are all of these small problems that you have to solve. I learned more from doing that project than doing any other project. I learned how to fund a large, speculative research project; How to navigate working with people from different perspectives and disciplines; How to get institutional buy-in and support; and finally, how to communicate something that is really novel to researchers.

For me, it felt more like an entrepreneurship activity rather than a research paper because, while there is this big research component, so much of the problem involved convincing researchers who were not used to a machine learning approach. The research was very polarizing, some people thought it was a bad idea, and some really liked it. That always tells me that I’m on to something because I think that if you’re doing anything that is novel you should not expect everyone to agree with you. You are going to get a diversity of opinions and it is really up to you to trust yourself and develop that instinct about what you think will work. I trusted myself through that process and learned to hone that research intuition. I knew it was a good idea and even if it did not work out, I was proud of the work we were doing. But I think it did work out so I’m really proud of that paper and I’m proud that with research.

This publication opened up a lot of doors. Using satellite data is now standard for hedge funds who use this information to develop trading strategies. In my own work, I’m not using satellites not just to look at poverty but also to look at conflict such as bombing damage in Syria, I’m looking at population growth and how to evaluate it–so that’s opened up another path of research that I think is really cool.

What is the most important lesson you hope students walk away from your classes with?

The most important lesson I want students to take away from my class is that there is a big difference between knowledge and information. Data is just a deluge of numbers, and we live in this data rich world, but information that a company can take action on is still very scarce. I try to convey to my students practical, hard technical skills, that help them separate out information from the data, the kind of which any firm or business could take action on. This requires, not just technical skills, but critical thinking skills to know what to do with the analysis. Students can incorporate of all of the critical thinking skills that they’ve learned in other classes and apply it here. Maybe in literature class and they’ve learned critical thinking skills; maybe they have taken a science class and they’ve learned critical thinking skills there. They can incorporate and use all those skills and apply them to a different domain. So I think I want my students to come out learning some technical skills but I want them not to just be able to learn formulas, I want them to think seriously about what’s happening and using critical thinking just in this different domain of data and statistics.

What makes your class different and/or why should students take your classes?

What’s different about my class is that you are going to learn things in my class that’ll be useful in a variety of domains, regardless of which direction your career takes you. If you go into finance or marketing, I think you’ll find something useful. Data and machine learning are skills that employers really want students to come out of school having. I am not teaching skills that aren’t only useful 20 years out when you’re a manager but rather skills you can apply tomorrow in your internship or on the job.

I understand the material can be dry or a lot to absorb if you’re new, so I try to make class fun. This is not a kind of course that just purely dry lecture. We tell jokes, some memes appear in my slides. And I show examples of people who use data and machine learning to do something extraordinary or exciting that students might not realize is possible.

What’s the accomplishment you’re most proud of in your career or what goal are you working toward?

I am most proud that I am successfully able to collaborate with researchers from a variety of domains; not just economists but also computer scientists, political scientists, statisticians and business people. I’m proud that I can “play well with others.” One achievement that I’m most proud of is that I had a paper accepted at a conference called NeurIPS which is a leading computer science conference. I’m an economist who always wished he could also be a computer scientist. The fact that my research was accepted by computer scientists was incredibly fulfilling.