With much of the country battered by blizzards, and students racking up multiple snow days, we turned to expert David Sukhin—inventor of the Snow Day Calculator—for some forecasts on when it might all end.
Sukhin, a junior at MIT, majoring in computer science and management, first came up with the Snow Day Calculator idea when he was in sixth grade. At first, many data points needed to be entered manually for snow day predictions. But by the time Sukhin was 16, he’d made the calculator more automated. In 2012, he added a 99-cent Snow Day Calculator app, in addition to the free online version.
I caught up with David in the middle of a New England storm to find out if his Snow Day Calculator can predict when we can all put our shovels down and send our kids back to school.
How did you come up with the idea for the calculator?
I’m not really sure; I was just kind of messing around. I would see that it was going to snow, and I thought, “Can I predict if we get a snow day?” At first I started it just for my school. Eventually, I generalized it and made it more applicable to everywhere. Now, people just type in their zip code (in the U.S.) and they get the same predictions I was trying to make back when I started.
What data does the Snow Calculator access, or is that your “secret sauce”?
Well, there is some secret sauce in there, but when you type in your zip code it contacts with weather.gov, which gives an hour-by-hour forecast. This includes how much snow is going to fall, how strong it is, if it’s freezing rain or snow or sleet, and what is the chance of the precipitation. All of that data then gets crunched with the secret sauce and you get the core prediction. Then the extra factors such as how fast certain areas clean up the snow, how tolerant areas are to lots of snow, type of school (public, private), and whether it’s a rural or urban area gets mixed together to come up with the percentage chance of a snow day.
How accurate are the predictions?
The predictions haven’t been wrong for my high school and middle school! The accuracy is what keeps people coming back, so I make sure the calculator is as accurate as possible everywhere. The Snow Day Calculator is also constantly learning about the results of its predictions to becomes even more accurate.
How are different geographic areas affected differently?
Certain areas—like Georgia which was immobilized by two inches of snow last year—don’t have the plows and equipment to clear the snow, so they are more likely to have a snow day with only a small amount of snow. I try to take that into consideration with the calculator.
What inputs do you look at to help you make the predictions?
The Snow Day Calculator only relies on three inputs now – zip code, snow days this year, and type of school. The manual version has some additional number inputs to play with.
Public vs. private school has its own weight (and rural public and urban public are also options), and the number of snow days so far this year is another factor. Administrators get a little more reluctant to call a snow day when there have already been several, because no one wants to be going to school in July!
When a town has had a recent snow day, the calculator looks up what the weather was that day and it compares it to the forecast, and that is factored in as well – the program learns from the previous data. It is interesting because administrators want to be cautious too – no one wants to be responsible for kids being stuck in school or accidents that may happen, but they need to balance that with the required school days.
Could there be a spin-off calculator geared for school administrators in the future?
That’s actually something I’ve been thinking about. I recently met someone who knew the person responsible for calling off classes at Boston University. They said that BU was planning on cancelling, but first wanted to see what Boston College was going to do – no one wants to be the first one.
I’ve been looking into a service for the administrator crowd, having them sign up to get in-depth weather information and timing of a storm. It would give them the percentage chance of a snow day and might also broadcast when an administrator first decides to cancel school, in Boston for example, to other area administrators before it is publicly broadcast.
What’s the most fun or interesting part of this project?
This has been a super fun project, and it opens doors to cool new things to try. The best part of the job is that people send in comments and suggestions, and it’s really interesting to see all of those.
Do you have a mentor that has helped guide you with this?
I showed the design to my family and friends as I was developing it, and they gave me their input which was very valuable. And my uncle is in the tech space and has helped me with the servers and capacity and getting a full-scale web application up and running with thousands of people hitting it. But, the math part sort of grew with me. As I learned more math skills, I would apply them to the Snow Day Calculator to improve it, and I will continue to do that.
What do you want to do after graduation and is it Snow Day Calculator-related?
The calculator is a great project that I’ve been working on since sixth grade, and I do see myself continuing to work on it in the foreseeable future. It’s fun and it’s a great testing ground for other projects because there are thousands of users and I can get feedback almost instantly. It’s also a great way to meet people, to talk to people, and to hear their ideas.
But, as far as after graduation, I’m exploring my options. I’ve been considering finance, maybe working on a start-up, or doing my own thing. Mathematical predictive analytics will probably be a big part of that.
Photo Credit: Boston Globe/Contributor