This summer, I worked with Professor Judd Kessler, who is part of Wharton’s Business Economics and Public Policy Department. During my work with Professor Kessler, I focused on data-driven public policy, which means we collected data to infer the effects of policy on different segments of the economy and find ways to optimize policies already in place. One example of the work we did over the summer is with Heifer International, a nonprofit working to eradicate poverty and hunger through community development. Heifer raises money by sending catalogs to prospective donors, including individuals, families, funds, schools, organizations, and more. My job was to take the data collected by Heifer and find what catalogs optimized donations from each demographic (i.e., schools, families). With my data science background, specifically in deep learning and machine learning, I decided to create an algorithm that would find what donation value each catalog would bring. Unfortunately, due to a lack of data on Heifer’s part, we could not continue with the project. 

For my next project, I worked in tandem with Teach For America to find what grants given to prospective teachers lead to the best returns. In this case, returns are measured in the length of the teacher’s tenure at the school. For this project, we needed to find what subjects had done during a specific period, specifically if they had taken part in education in some form. At first, this task was projected to take months due to the sheer number of subjects and the time it took to find their work experience online. As a result, I was asked to program an automated web scraper, specifically, the first academic LinkedIn web scraper. This task taught me a lot about the different technical methods available and how data is collected to create machine learning algorithms. I also learned a lot about working individually. There were many instances where the code was not working correctly, or when accounts I was using got blocked, and I had to find a way to get around a problem, independently. By learning how to face an open-ended question, and get around obstacles on my own, I feel better prepared for challenges I may face ahead, both in the classroom and after graduation. Finally, after a few weeks, I was able to use my web scraper, which boasts an 80% accuracy rate and has already decreased the amount of time RA’s spend looking for subjects online. 

I chose to research with Professor Kessler because I wanted to understand better how data can be used in economics, specifically macroeconomics. In class, we learn about macroeconomic theory without discussing the data and research behind the theorems. Now, because of the opportunities, I have had over the summer, and because of Professor Kessler’s incredible mentorship, I have a much better understanding of economic analysis and what questions in economics can be answered through data and how information is collected and used to make inferences that are later published.