Stats goes cool

On Friday I got to pre­tend I was a stu­dent again. I sat in on Auroop Ganguly’s grad­uate class, Applied Time Series and Spa­tial Sta­tis­tics for the second of two guest lec­tures on the sub­ject of fore­casting. Last time, it had to do with fore­casting the finan­cial impacts of nat­ural dis­as­ters like Hur­ri­cane Sandy, which I wrote about with the help of two of Ganguly’s stu­dents. This time the guest was Mike Liebson of Oracle, a hard­ware and soft­ware com­pany that designs advanced ana­lytics tools to deal with big data. The topic was fore­casting sales, demand and supply in the com­mer­cial industry.

At the begin­ning of the talk, Liebson showed two pic­tures of dapper young men and said “these guys made sta­tis­tics cool.”

Photos via Wiki­media Commons

The guy on the right you prob­ably rec­og­nize but may not know what he’s got to do with stats. The guy on the left may look less familiar but when I tell you his name you’ll know right away why he’s up there.

Brad Pitt por­trayed Billy Bean in the movie Mon­ey­ball, which tells the story of how the Oak­land A’s gen­eral man­ager revived the team on a budget using nothing but bat­ting aver­ages and other base­ball stats. Nate Silver is the sta­tis­ti­cian who cor­rectly pre­dicted the out­come of the recent pres­i­den­tial elec­tion in all 50 states. He’s also a keen saber­me­tri­cian in his own right.

Char­ac­ters like these two, Liebson said, have proven to the world that big sta­tis­tics is pow­erful stuff. He told his own story, in which a Dunkin Donuts exec insisted for years that the coffee and sweet treat mag­nate could keep America run­ning with the help of big data ana­lyt­ical tools like those pro­vided by Liebson’s employer. But ulti­mately he con­vinced her that this was simply the direc­tion the world is going and to stay on top of the market, DD would have to play ball.

In 1886, John Wana­maker, the father of the modern depart­ment store, said the fol­lowing: “Half the money I spend on adver­tising is wasted; the trouble is I don’t know which half.” DD, Liebson said, wanted to figure out the solu­tion to that seem­ingly impos­sible problem. They wanted to know whether an ad cam­paign pro­moting cool bev­er­ages in the sum­mer­time was actu­ally improving sales or if the increased num­bers were based purely on seasonality.

With mil­lions of cus­tomers across the globe and as many trans­ac­tions each day, crunching the data on this ques­tion is no easy task and requires some big time ana­lyt­ical tools. But it’s no longer impos­sible, because those tools are emerging from places like Oracle and SAP, their number one competitor.

The same tools allowed DD to figure out whether other prod­ucts were being “can­a­bal­ized” or expe­ri­ence “halo effects” due to the pro­mo­tion, as well. That is, they could see whether lemonade sales would go down because of a pro­mo­tion on coffee Coolattas, for example. Or if someone coming in for a Coolatta also left with a donut.

Big data isn’t just the messy unstruc­tured amal­ga­ma­tion of tweets and other social media stats that we tend to think of, Liebson said. It’s also the increasing scale of infor­ma­tion gen­er­ated as large com­pa­nies glob­alize and expand even fur­ther. While it seems like such a mind-​​boggling amount of data could never be prop­erly parsed, advanced tech­nolo­gies are making it pos­sible and actu­ally changing the way busi­ness is done.

Base­ball isn’t the only thing affected by big stats and cool sta­tis­ti­cians. It’s just the way of the world now.