MIDDLETOWN – On a Sunday morning in February, five High Technology High School students met at 16-year-old Andrew Eng’s house to start a math competition that blended data science, computer programing and critical thinking to model the future of remote work.
Once they clicked start, they had 14 hours to digest the question, conduct background research, brainstorm how to tackle the problem, model the issue at hand and submit a 14-page paper detailing their findings.
And they came prepared — with whiteboards, Dunkin’ Donuts and Costco Pizza to fuel themselves.
It must have worked. Out of more than 2,700 high school juniors and seniors who entered, they are one of just six teams nationwide heading to the final round, where they could win thousands of dollars.
“We’ve had smart people in the past, but you have to work as a team,” said Raymond Eng, a High Tech High teacher and the team’s coach, who is unrelated to Andrew. “You can’t let ego be an issue. So, it’s a team effort. It’s a huge problem … so, it has to be a team effort of five people.”
The competition is called the MathWorks Math Modeling Challenge, or M3 Challenge. It draws teams of students from all over the US and UK to compete for scholarships. The first-place team will receive $20,000, the second-place team will receive $15,000, the third-place team will receive $10,000 and the fourth-place team will receive $5,000.
Technical Computing Awards that range from $1,000 to $3,000 will also be awarded to teams with the best use of computer programming. The competition is a program by the Society for Industrial and Applied Mathematics (SIAM) and this is its 17th year.
The five-person team from High Technology High School in Middletown passed two rounds of assessment and are invited to the final round, an in-person presentation of their paper to a panel of professional mathematicians in New York City on April 25.
David Chang, 18, and Alexander Postovskiy, 17, the only seniors on the team, are returning finalists from last year’s competition.
“Last year, our reaction was ‘We’re definitely not going to make semi-finalist,’ then we made semi-finalist,” Postovskiy said. “And then ‘We’re definitely not going to make finalist,’ and then we made finalist and then ‘We’re definitely not placing,’ … and then we got second.”
This year, Kevin Guan, 17, said the team thought their paper had a chance to make the semi-finalist round, but not the finalist round. They were surprised to be invited to the finalist round.
Raymond Eng, a former engineer with a doctorate who worked at nuclear power facilities, said he has encouraged students to compete in M3 since its inception. In the past 16 competitions, students have won scholarships in 13 of them. He teaches multivariable calculus and engineering design and development at High Technology High School.
The pandemic forced last year’s team into a 14-hour conference call to model solutions to the digital divide — the issue where some people have less access to high-speed internet than other. But this year the new team got to work together in person.
‘We need our students to be in school’: Manchester school chief targets COVID learning gaps
Breaking down the problem
Andrew Eng said the team spent the first half hour to an hour brainstorming how they would tackle the problem. The team then split. Postovskiy and Ivan Wong, 16, were assigned to the first part, discovering the number of remote-ready jobs in each industry in 2024 and 2027 in a number of cities.
Chang and Guan were tasked with the second part, looking at the tendency that certain people would choose to work remotely based on their age, sex, occupation, educational level, spousal employment and the presence of children and elderly family members. Eng started working on the first part before moving to part two when the second group began experiencing difficulty.
Guan said the team was held up trying to classify people into categories.
“In machine learning there’s two types of models that you can use. One is a regression, which generates a number, and one is a classification, which basically sorts into a category,” he said. “The night before, when I was looking at this, the only code I had done was with regression, with generating numbers. But this problem wanted a category, which the code I was originally using didn’t work for. So, I spent like an hour and a half, two hours just figuring out how to use another coding language to do the categorization part.”
Even with slight hiccups, the team was making good progress until the third part. That’s when everyone worked together to figure out the economic benefit remote work could have in the cities they studied.
“We got a little complacent and then toward the end we were like, ‘We should have spent our time better,’” Guan said. “The last few hours were a kind of a scramble. … (Last year’s team) had a similar experience … where they were literally editing the doc as they were submitting it, and we were like laughing at that. And then it happened to us.”
Postovskiy said, “I think we finished the part three model like 10 minutes before the 14 hours.”
He said the highlight of the competition for him was realizing in part one that some of their figures could come up with a negative probability, which wouldn’t work for their model. “So, we spent an hour sitting on the floor with these three whiteboards trying to interpret it and trying to do linear algebra to make it make sense.”
More than just math
While everyone on the team has a strong math background, to succeed in the M3 Challenge, competitors must have experience with the competition itself.
“The knowledge of how to do M3 just passes down by practice and like orally through the years as the teams are mixed,” Postovskiy said.
Andrew Eng said, “It was super helpful to have two very experienced seniors. Without them, I think (the) juniors would be extremely lost at where to even go and how to approach these problems.”
Wong said a lot of the learning with the M3 Challenge is self-study, just thinking critically as they went.
“You don’t necessarily have to be a very good mathematician. It also involves creativity and teamwork. How you define the problem,” he said. “We’re trying to qualify impact. So, there’s a lot of ways to codify impact. So, you just have to creatively find a way that you want to think about impact on the cities.”
Raymond Eng said, when choosing students for the team each year, he had potential team members write a three-page paper on the pros and cons of a social issue of his choosing. He gave an example of free community college.
“There are people who have very good math credentials, but this is not a math competition per se,” he said. “Making a model with a great amount of uncertainty involved and justifying, articulating. So, it’s communications. So, you need a balance.”
Postovskiy said the judges are really looking for the logic behind the models they present to them.
He said “It’s moreso the reasoning you use to produce your model is both mathematically sound and showed that you put a lot of thought into the factors that affect something in the real world.”
Last year’s competition was a Zoom affair, in which teams had a 30-minute time slot to log in together and present.
Postovskiy said he was star-struck last year when he Googled a judge who was asking questions about the former team’s model. They had used a technique called integer programing. Postovskiy said he found out that the judge was one of inventors of MATLAB, a programming language.
“We have these judges that are really well respected and have done a lot in the field,” he said.
Raymond Eng echoed that sentiment. “This is a golden opportunity for the students.”
He added, “It’s a real learning experience and (the competition’s problems are) real life problems where there is no definite answer. It’s a matter of, you want to be objective about it, you want to do the pros and cons. And look at it objectively. I thought this was a great opportunity, a real-life opportunity for what is it in the real world.”
Chang said he is looking to study computer science and astrophysics in college and Postovskiy said he wants to be a math major.
Wong said he is looking at studying computer science and math; Guan said he is considering some sort of interdisciplinary study that mixes computer science with another field; and Eng said he is considering the intersection of data science and business.
According to the M3 Challenge, the other finalist teams are from Mequon, Wisconsin; Osprey, Florida; Lincolnshire, Illinois; Winnetka, Illinois; and Huntsville, Alabama. The top ranking UK team from Watford, England, has also been invited to present their paper at the final event, although they will not be competing.
Olivia Liu is a reporter covering transportation, Red Bank and western Monmouth County. She can be reached at [email protected]
This article originally appeared on Asbury Park Press: M3 Challenge finals reached by High Technology NJ high school students