In the midst of March Madness, Stanford University biomechanical engineering student Muthu Alagappan recently presented his ideas at the MIT Sloan Sports Analytics Conference on how statistics could determine a player's position - and transform those positions - based on mathematical filtering of per-minute production. Rather than merely five positions, though, he identified 13.
As The New York Times explains:
... traditional positions may be convenient and familiar, but the players grouped as power forwards or shooting guards have become so diverse that the terms themselves have lost all meaning ...
Point guards are expected to pile up assists. Centers are expected to rack up rebounds. Basic statistical templates exist to reinforce the prototype at each position, and those who embrace unique styles of play often find themselves on the receiving end of veiled value judgments. ... Alagappan's model finds a place for those players by making the relationship between counting stats and positions both more explicit and more accommodating.
It's just another example of the potential for analytics to transform the world as we know it - and that statistics have gone from geeky to a chic field. That's a point made in an Information Management piece about the casting of Brad Pitt as Billy Beane, the Oakland A's general manager in the 2011 movie "Moneyball," who used analytics to transform the team.
The author, Elissa Fink, chief marketing officer for Tableau Software, offers three ways to address the shortage of talent trained to effectively use analytics:
I wrote last week about CITO Research's position that the answer is to grow your own data scientists. Fink's role, obviously, is to sell software, so she advocates leaning on technology. In an interview last year, though, Jack Phillips, CEO of the research firm International Institute for Analytics, told me he believes that as technology grows more sophisticated, it increases the need for analytics training to know the right questions to ask.
In an interview at SmartDataCollective, Gregory Piatetsky-Shapiro, president of the consultancy KDnuggets, makes an important point. (He dislikes the term "data scientist" and calls the work these folks do "knowledge discovery," hence the "KD" in his company's name. In the article he also talks about why he thinks Sherlock Holmes would be good at this work.) Said Piatetsky-Shapiro:
I coined the term "knowledge discovery" to emphasize that what we want to find in data is some understandable knowledge and not just incomprehensible patterns. Much of the big data analysis and many of the best predictive algorithms, unfortunately, produce results that are incomprehensible. So we have a tension between accuracy and understandability, but I think that better understanding of predictive models will contribute to increased trust for such models, and may also help resolve some privacy issues by providing increased transparency.