IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

New York Schools Apply Machine Learning to Study Inequality

Five students at Shaker High School, New York are finalists in a statewide competition for their report that used World Bank data and machine learning models to assess the impacts of various factors on income inequality.

AI+machine+learning+city.jpg
(TNS) — A group of North Colonie students is using machine learning to understand how factors like education, the environment, demographics, and infectious disease are tied to a country's level of income inequality.

The five Shaker High School students are finalists in the statewide High School Fed Challenge for their research and analysis, which yielded a report that will be among 12 papers to be published in the inaugural issue of the Journal of Future Economists this summer.

Their report uses World Bank data from around the world. Their findings are especially timely in a year when virtually every nation has been engulfed by the pandemic.

"Our main finding was that infectious diseases had the (greatest) effect on inequality. And all this data is pre-COVID-19, so it was very interesting," 11th grader Avi Bagchi said. "We know that the coronavirus was a major cause of inequality and really has exposed inequality in the United States."

The annual state contest, which is run by the New York Federal Reserve Bank, has evolved from an in-person competition to an academic paper challenge in which student teams research and analyze an economic theme.

"We were blown away by the quality of the 66 papers we received," said Heather Daly, director of education programs at the New York Fed. "Just as professional economists produce a mix of qualitative and quantitative analyses, so did the students. Every team that participated did truly outstanding work."

The High School Fed Challenge aims to encourage high school students to learn more about economics and promote economics as a subject for study and a career possibility.

Bagchi and his co-authors — junior Harrison Fazzone and 10th graders Sahana Vinothkumar, Priya Musuku and Roshni Ramesh — are members of their high school's History, Research, and Competition Club.

The club, which started out as a group of friends who competed in the annual Academic World Quest, an international affairs competition, expanded during the pandemic to take on other kinds of contests and challenges. In the last year, its membership has grown to more than 80 members.

The students who signed up for the Fed Challenge were self-motivated, collaborative, and rarely asked for guidance, according to Shaker High School social studies teacher and club adviser Andrew Krakat said.

"I was very impressed with the work this group is doing, especially the higher-level statistical analysis, pulling out the data points beyond GDP (gross domestic product) because a lot of inequality reports are focused on using GDP as an economic measure," Krakat said. "They identify that there are actually these other measures that are much more predictive of economic inequality."

The team was unable to meet on school grounds since the pandemic has divided students attending class part-time into different cohorts, and some students are fully remote. Instead, the research happened on Google meets, which was an awkward adjustment at first, according to 10th grader Vinothkumar.

"It was definitely hard to adjust from a normal in-person collaborative where you bounce ideas off of each other," Vinothkumar said. "We would have late-night ideas and text them in the chat and we would look at them at various hours. It was so wild, but so productive at the same time ... we weren't limited to the boundaries (of set club hours) we would normally be."

Students said they were driven by a spirit of competitiveness, but also their passion for the topic of wealth inequality.

"With the pandemic, we see how the workforce is really important to shaping how conveniently and easily we can live our lives," Ramesh said. "Income inequality we see is such a big barrier to medical care and life expectancy and we felt like really exploring this issue is something that we needed to do."

Harrison and Bagchi came up with the idea after Bagchi used machine-learning in another class to analyze invasive species. They used Python programming to collect and organize the data. The data was then run through statistical models that looked at how various indicators impacted a country's income inequality, quantified by a measure known as the "Gini coefficient."

Vinothkumar, Ramesh, and Musuku helped with the analysis and pulled the information together into a cohesive report. Another strong indicator of income inequality was the environment; rural communities tended to have greater inequality than urban areas, according to the findings.

"It was very interesting to see that non-income factors and non-financial factors contributed to wealth inequality. That was what really struck me. it was a bit unusual but it did make a lot of sense when you think about it," Musuku said.

(c)2021 the Times Union (Albany, N.Y.). Distributed by Tribune Content Agency, LLC.