- Diabetes in the Digital Age
- Diets and Data: Counting More Than Just Calories
- Why You Can No Longer Blame it On The Weather
- Your New Personal Shopping Assistant
- IBM Watson Vs. Cancer: The Race Is On
- Click Here to Reinvent Insurance for the 21st Century
- A 3-D Vision for Education 2.0
We human beings have an insistent need to believe that there are reasons for the things we do, that our behaviors are somehow steered by functioning and reliable rudders. But sometimes our rudders lie too deep in murky waters to observe and analyze their operation accurately. When we are confronted with any perplexing trend in human behavior, our instinct is to attribute the pattern to some erratic underlying variable. Sometimes that variable is as simple as the weather.
You might be surprised to learn how many people blame it on the rain. Last July, for example, when the New York Mets turned around a dismal season, their manager, Terry Collins, actually explained that a spell of hot weather was responsible for changing the physical properties of the baseballs in a way that drove up the number of runs at their home field.
Perhaps he was borrowing a play from Janet Yellen, Federal Reserve chair, who two months earlier had explained away a disappointing report on the nation’s gross domestic product by lamenting the “unusually cold and snowy winter weather” that had been a factor in the GDP’s slowing growth.
But if you’re going to cast blame (or give credit), then you need the data to back it up. When Milli Vanilli recorded its lyrics in 1988, our weather forecasts were still pretty unreliable. Today, however, in many parts of the country, we can record temperature, humidity, and barometric pressure at a remarkably fine resolution, in no small part owing to an explosion of new WiFi-connected, sensor-laden devices. This expanding infrastructure has not only refined our ability to accurately forecast the weather but also, when combined with other data, has provided industries with insights into how weather changes human behavior. For businesses, these insights can lead to substantial boosts in revenue.
According to weather experts, the sources we rely on for our weather-related data have been steadily decentralizing over the past decade and, in doing so, have filled major gaps in our forecast data. The National Weather Service, a branch of the National Oceanic and Atmospheric Administration (NOAA) that has been collecting valuable data from satellites, radar, and high-altitude balloons for many years, is grabbing more and more of its readings from sensors outside its own network.
“The non-NOAA list is huge,” says Robert Jacob, a computational climate scientist at the Argonne National Laboratory. Through a project called the Array of Things, Jacob is spearheading an initiative to install weather-related sensors on streetlights around Chicago. “It’s things like stations run by universities, stations run by state transportation offices. It’s already an incredibly decentralized operation.”
Hobbyists and weather fanatics are adding to the data deluge with efforts coordinated through groups like the Community Collaborative Rain, Hail and Snow Network (known as CoCoRaHS4) and the mPING smartphone app, which invite ordinary citizens to report precipitation events along with their precise location data. Some of the contributions are much more sophisticated. As the cost of sensors and Internet service has come down, more people have started putting up comprehensive weather stations in their backyards and are relaying their data to meteorologists.
According to Mark Gildersleeve, president of the professional division of the Weather Company, the number of private stations that inform their forecasts jumped from 30,000 to 134,000 in the past year alone.
As temperature and humidity sensors get slapped onto more and more consumer devices, even the idea of what constitutes a weather station is beginning to change. Is a smartphone a weather station? Is a car a weather station? Is a solar panel a weather station? Increasingly, the answer is yes.
For example, “imagine that we have data from a connected car that tells us the windshield wipers are on,” says Gildersleeve. “That’s a real good indicator that it’s raining at that precise location at that precise time.”
By tapping into these diverse sources, services like the Weather Company can now achieve a remarkable level of specificity in their forecasts. “They now generate 15 billion weather forecasts a day, and that number has been increasing because what they’ve been doing is they’ve been improving their stochastic science ... down to a nine-digit ZIP code in every major market around the world,” says Glenn Finch, global leader of Big Data and analytics at IBM.
For Finch, it really gets interesting when you combine this precisely localized understanding of historical weather events with other data and use that to make better decisions. Institutions of all kinds—militaries, state governments, retailers—have something to gain, says Finch.
Consider a café owner who is trying to decide at which point in the season to shift the café’s promotion strategy from lattes to iced coffees. The owner, of course, knows that as summer approaches, more people will want to drink iced coffee. But he probably doesn’t know exactly when to pull the trigger, and so he guesses and hopes he’s right.
Through a partnership with the Weather Company, IBM recently analyzed the sales data for one of its clients and found that consumer behavior was remarkably reliable. As soon as ambient temperature hit 69 degrees, according to Finch, people started to prefer iced coffee.
“The client then had the ability to augment a promotion strategy based on weather, instead of thinking ‘She just doesn’t like lattes anymore.’”
In another example, IBM used data from the Weather Company to help a pharmacy owner decide on the best time to stock up on antihistamines. Allergy season varies dramatically around the country. With advanced knowledge of local weather and pollen counts, a retailer can avoid buying and promoting products too early in the season—causing a needless loss in revenue.
“Two months later, I’ve got to sell it at a discount, and I take the hit in margin. You have no idea how significant that cycle is. If you’re off by a month, it can be huge,” says Finch. “What few people understand is the highly connected relationship between the retailer and the consumer-products manufacturer and how important that timing actually is.”
Even so, Finch has encountered a perplexing amount of resistance among many clients to take weather into account. “So there is a massive change-management aspect to this whole thing that I think everybody underestimated,” he says. “What we’re doing is we’re unlocking a paradigm shift.”∎