With flurry of data, this app is a snow day predictor
It sounds like a gimmick Bart Simpson would dream up, but the “Snow Day Calculator” is real. And even though many grown-ups don’t know it exists, when a storm’s coming, the predictive tool is the talk of students everywhere.
Like so many inventions, the calculator— which started out online and is now an app, too — was born of personal need, specifically the desire of New Jersey middle-schooler David Sukhin to know when snowfall would prompt a school cancellation, letting him put off his homework.
“My friends and I would all agonize about whether school would be closed,” he said. “I wanted to use all the data and algorithms I possibly could” to get the answer.
That was nine years ago, and, as you’d guess, a middle-schooler who thinks about data and algorithms definitely applies to MIT. Sukhin is now a senior there, studying computer science and business.
Sukhin has programmed his algorithm to use the data points you’d expect: National Weather Service predictions about the amount and type of precipitation expected; the timing of the storm; and historical information about what it usually takes for a district to cancel school.
But he uses squishier measures, too, including a storm’s social media and TV buzz.
“Hype can swing the [likelihood of a snow day] by 10 to 15 percent,” said the young man, who claims to have amassed more snow-related school data than any person living.
Sukhin’s calculator is simple to use. Kids input a few pieces of info: ZIP code, the number of snow days to date, and the type of school (urban public, rural public, boarding, etc.).
The calculator then returns the odds of an impending snow day declaration.
If “Limited” comes up — reflecting a zero to 55 percent likelihood of a snow day — it means students better be ready for school the next morning.
But a “whoo-hoo!” signals 87 percent to 99 percent odds there will be no school or an early dismissal.
“I feel like I’m gambling,” said Ben Kremer, 16, a sophomore at Brookline High, of the times he uses the calculator. “I always hope the number is going to be really high.”
Sukhin doesn’t know his batting average for accurate predictions, but he says since 2012 — when he started at MIT — he’s been wrong about cancellations in Greater Boston only once or twice.
“I have a nine-year streak of accuracy at Valley View,” he said of the New Jersey school that started it all.
The online calculator is free to use, but Sukhin sells advertising space and subscriptions — a few dollars buys a winter’s worth of text alerts for closings up to three days in advance. And over the past several years, about 60,000 people have bought one of his 99-cent apps.
Business is brisk. During a storm, the calculator typically generates between 1 and 2 million predictions, its creator said, noting that some students — and teachers — return multiple times, hoping for a better outcome (and some parents visit, too, hoping for their own version of a good day).
Boston, by the way, ranks somewhere in the middle on the winter-toughness scale. Given equal amounts of snow, we’re more likely to cancel than are schools in Michigan, upstate New York, and northern New England, but less likely to cancel than those in Washington, D.C., or Texas.
Our snow-day twin: New Jersey, he said.
Sukhin doesn’t have the power to cancel school, but some users consider him omnipotent. “Can you make it snow?” one kid e-mailed.
The answer is no, but his people don’t hold it against him. “There’s nothing he can do if it’s not going to be a snow day,” said Jonah Rothman, 11, a sixth-grader at Charles River School. “It’s the weather’s fault.”
Well, mainly. School or other officials do make the ultimate decision. And they are well aware of the stakes.
On Sunday night, with Sukhin’s calculator predicting an 89 percent chance of cancellation, Newton’s superintendent of schools, David Fleishman, released an automated call alerting families there would be no school Monday.
Said Fleishman: “I was told people cheered for me at a Super Bowl party.”