In August, more than 700 teams comprised of statisticians, programmers, engineers and students from across North America applied to enter this year’s “2018 NBA Hackathon” — a data-driven competition to “build tools that solve important and challenging problems in the NBA.”
It was then that a team of three mathematics students from NJIT applied for and earned distinguished selection into the contest’s final 20-team field, scheduled to compete at the NBA headquarters in Secaucus, NJ this past fall.
By the end of the grueling 12-hour event, “Team NJIT’s” Bayan Mafarjeh ’18, Raiha Khan ’19 and Jean Sebastien Darius ’19 emerged 2018 NBA Hackathon finalists, taking home the competition’s second-place prize.
“It was crazy,” said Mafarjeh. “We were going up against Ivy League teams and had to tackle tough problems on the spot, so I couldn’t have predicted we would get as far as we did. I really wanted to brand our team name ‘Team NJIT’ so that everyone there would recognize the university and the work we were doing."
"The whole experience opened a lot of doors, and since the hackathon I have been in ongoing interviews with the Detroit Pistons and Oklahoma City Thunder," added Mafarjeh. "We were able to network with a lot of great people we would never have otherwise, like commissioner Adam Silver and deputy commissioner Mark Tatum.”
“I even got to ask commissioner Silver why he let Kevin Durant go to the Golden State Warriors,” joked Darius. “Honestly, my main goal has been to go into economics and graduate school, but after this, I could consider a detour into sports analytics.”
“Seeing first-hand the way data analytics is growing and being applied on both the basketball and business side was really cool,” added Khan, a student in Albert Dorman Honors College. “I really hope that our success and experience can attract the attention of more people from NJIT in the future and make them aware of this kind of event.”
This year’s 2018 NBA Hackathon was split into two tracks, each featuring ten competitive teams — a “basketball operations” track and a “business analytics” track.
Khan, Mafarjeh and Darius were accepted into the competition’s “business analytics” track for their summer project proposal involving data modeling that could be used by NBA teams to boost overall ticket revenue. In their proposal, the team developed a computational model to analyze secondary ticket sale data from the 2017 NBA season on StubHub.
Using their model, the team was able to predict how much each team could benefit from either reducing or increasing the availability of their season tickets, which are often incentivized at a reduced rate before the start of each season.
“Essentially, we went through last season’s data and asked, ‘If a given NBA team capped their season ticket sales at a certain limit, would this team have done better or worse in terms of overall revenue?’” said Mafarjeh, who spent three years as a member of NJIT’s men’s soccer team. “There are obvious examples our model showed like the Golden State Warriors, where the team would benefit from capping season tickets and selling more tickets on a per-game basis since they are so popular and will sell out each game regardless. On the other hand, the Nets barely have their arena filled, so they need to offer more season tickets with the incentive of discounted rates in order for people to come to games.”
“When we first wrote our Python script to analyze the StubHub data, it took about 10 hours to run everything and get our results,” said Darius. “We used a lot of what we learned at NJIT to start the project, but then we had to do a lot of research on our own. Eventually, we were able to get the runtime to just three minutes.”
At the event, the team first impressed judges during a three-hour timed “sentiment analysis” challenge, which involved text mining social media data to rate the social media popularity of NBA players. With the clock ticking, the team wrote code and developed a method for ranking the popularity of Utah Jazz players based on positive, negative and neutral user comments posted by the Utah Jazz Reddit community over the course of last year’s 82-game season.
“It was like a science fair with four judges and a media crew with cameras going to different team’s tables,” said Khan. “When we saw them walking around to different tables along with their camera crew it was pretty nerve-wracking.”
“Jean is good at writing Python code, Raiha is good at natural language processing and I was analyzing quality scores of players, said Mafarjeh. “It was hectic, but we used all of our strengths and worked together.”
In the competition’s final challenge, Khan, Mafarjeh and Darius formally presented their NBA ticket sale data project before a judge’s panel of data scientists and league officials.
“We only had about 10 minutes to present, so we really wanted to make sure that we could get across our results to the judges and show them the future impact that this analysis could make on the company and for every team,” said Khan. “By the end, they were really nodding their heads.”
Upon being awarded for their second-place finish, Khan, Mafarjeh and Darius received a basketball signed by each of last year’s NBA Dunk Contest contestants, a $400 gift card to the NBA online store, and perhaps best of all, three tickets to any NBA game of their choosing.
“We are looking forward to those tickets, I already know I want to see the L.A. Lakers,” said Mafarjeh.
“The tickets are what we are in for now,” agreed Khan. “If we get to pick anywhere, I’ll fly myself out to L.A.”