Some time ago Seth Grimes interviewed me as par of his research on a paper called What You Can Learn From
Operations Research that Information Week has just made available. It's a nice overview and worth a read. In it he quotes me:
"Unless BI becomes both decision-centric — focusing on decisions to be made not on data to be stored and regurgitated — and more focused on operational decision making — the kinds of decisions that change transactional outcomes — then it is going to remain separate from OR."This is something I feel strongly about and I worry that "traditional" BI folks are not learning from OR/Data Mining/Predictive Analytics or at least that they are not learning fast enough. There is some case for optimism with papers like this one from Kurt Schlegel at Gartner talking about the need for BI to become more decision-centric but it is taking too long. OR and the techniques within that broad definition are very decision-centric, BI needs to learn...
Subscribe to this blog by RSS
From my experience, it's hard for an untrained user to even picture what things like optimization, analytics and machine learning can do. And even if they have a general idea, it's even harder to figure out how to get started.
Since you are asking BI people to use more of these techniques, I'm guessing that you are talking to people who are not trained to use them.
So is this not really a matter of attracting people with the right skills to work on problems posed by BI?
If so, then the first steps are to train these people how to pose their questions in ways that operations researchers can use. And to teach them to manage their expectations: that they can't expect a quick answer to every question and that they must work to help researchers as much as researchers work to help them.