Types of Data that could “win”

Recently, at Minnebar 2014, I attended a session by a Colin Tuggle and Joe Hebl called The Real Value of Big Data. In the talk they enumerated the following types of analytical data:

  • Descriptive – what is happening/has happened
  • Diagnostic – why does/did this happen
  • Predictive – what will happen
  • Prescriptive – what should happen
  • Excellent distinctions and applicable to not just “Big Data” situations. Of these four, it is most seductive to make assumptions based on Prescriptive data. If you are able to clearly delineate between Precsriptive and Predictive, however, Predictive is, by definition, a much more reliable data set upon which to make predictions.
    Additionally, it seems that incorrect data in Descriptive would lead to inaccuracies in the other data types. As a basketball fan, there is a new reliance in the NBA on what is being called “advanced statistics”. However, in this article Stan Van Gundy expresses that he doesn’t trust some of the stats. Saying, “I read a thing in the playoffs last year that said that New York isolated like 17 percent of the time….I’m watching their games, they isolate half of the time, at least.”

    Here’s a link to the PPT from the presentation by Tuggle and Hebl.

    “Data Wins” – an integral piece of my work philosophy

    I’m not sure where I first heard of it, but there was a philosophy of “Data Wins” with which I crossed paths not too long ago. It has been a driving force in my decision making. I believe it was an IBM philosophy, but a few cursory Googlings have turned up nothing to support this supposition. If anyone out there has any more information on where this came from, I’d love to hear more about its origins and tales of its practical implementation.
    In essence, the idea was that you want people who are passionate and care deeply about their work, but being emotionally charged about a subject can make it difficult(impossible?) to stay objective. So when data is available it’s findings take precedent over any team member’s feelings. Data ALWAYS wins should perhaps be the saying.
    For instance, a member of the team considers a certain feature to be paramount to the success of the project. However, the engineering time to implement the feature is prohibitive and polling of the target demographic doesn’t find the feature necessary. Although the team member’s feeling towards the feature are relevant, the data available now suggests that the feature is not necessary. At least in the MVP. Perhaps not even in Gold. This is a compelling and effective way to manage things, but it must be introduced to the team early on.
    Initially I hadn’t planned it as such, but each paragraph of this entry started with a different word that, in turn, started with the letter “i”.
    Interesting, I think.