This Wilmington man knows when you want a PBR

Steve Poulin has been working with SPSS for almost 18 years now. Some might say he’s a predictive analytics wizard.
You might be familiar with SPSS if you’ve ever taken a statistics course. The statistical software has been around since 1968 — yes, 1968 — making it nearly 50 years old.
“We’ll call it mature,” said Poulin, who works for Brea, Calif.-based software company Aviana Global. Poulin, who works out of the coIN Loft in downtown Wilmington, is the only East Coast employee. He originally hails from academia.

There's a lot of seasonality with beer consumption. You find those patterns of how beer demand has changed over weeks and months, and you forecast that.

Up until SPSS was acquired by IBM in 2010, Poulin was teaching at the University of Pennsylvania while working as an independent contractor with SPSS.
“Much to my surprise, I was an IBM employee,” Poulin said of the acquisition. “I went from Penn and academics to a corporate world.”
Poulin became an SPSS trainer with IBM for three years, until IBM decided to stop offering software training in 2013. So, in January of 2014, he became part of Aviana’s SPSS Practice Team as a senior consultant working in predictive analytics — or, analyzing data to make an educated prediction about the future.
“The essence of it is, you find patterns in the past and detect them for the future,” he said. “You can predict sales levels and then identify which customers are most likely to buy.”
Or, as a predictive analytics consultant, you can do what Poulin has been doing for Pabst Blue Ribbon for the past year-and-a-half: forecast beer consumption levels.
“What I did with SPSS is I built a routine that actually makes forecasts for specific distributors and products,” Poulin said. The total number of combination of distributors and products is 10,000, so Poulin’s routine makes 10,000 customized forecasts. What has he found?
“There’s a lot of seasonality with beer consumption. You find those patterns of how beer demand has changed over weeks and months, and you forecast that,” Poulin said.
Ideally, this forecasting routine works best with fresh data. So every week Pabst reels in another compendium of numbers, Poulin goes in and adds them to the loop.
But Poulin’s work interests go well beyond beer consumption. He wants to use this technology with Wilmington crime data to predict crime hotspots. He’s done it with other cities.
“I’d like to see it here,” he said. “They have the crime data.” Poulin said all you need for a hotspot prediction is time-of-arrest data and location. “Once you have that information you can make predictions of when crimes are happening and where, then you predict that into the future.”
Forecasting beer consumption is definitely cool. But this? This could actually make a real difference in Wilmington.
“It’s been done in other cities,” Poulin said. “I’d like to be involved in some way with this issue.”