Directions Magazine has a lot of expertise and experience in business geographics and related topics. I have just about none. I had the chance to get the “short course” at our Location Technologies and Business Intelligence Conference.
Michael Gonzales, longtime Business Intelligence (BI) consultant, planned to tell us about GIS and BI, but changed his mind and instead presented a hands-on experience to introduce us “GIS people” to what “real” BI is. Gonzales is a captivating and knowledgeable speaker, who challenged us to take a broader look at BI in general and the future of its relationship with spatial in particular.
Gonzales has a few key points that run through his argument:
He feels it’s BI architects that are ultimately responsible for getting spatial (or just the best tools) to their clients. Simple online analytical processing (OLAP) and queries are not BI. BI, he said, more than once, is the ability to provide the best possible information to make a decision. Most BI implementations, he suggests, do not do that. Instead, they rely on what they know: OLAP, SAS, pie chartsЕ These, he went on do not provide actionable insights, which they should.
Part of the problem, he maintains, is that OLAP was designed to be an interrogation tool, one used by subject matter experts. Instead, its results are being offered to non-subject matter experts (the 20 year old with the grease stains from his hamburger on his shirt who picks up the phone at a credit card company, for example).
With that in mind we looked at two BI solutions. One, Microsoft’s Analysis Services is part of SQL*Server. It owns some 25% of the market! And, says Gonzales, it’s not bad. We used its built in tools to develop a decision tree that helped predict the type of root a mushroom would have. While that’s not a fascinated topic or data set, the underlying idea is quite powerful. What factors can we use from the dataset to best predict nature of the root type? It was pretty impressive how far one could get, knowing very little.
A second solution, Polyvista, actually sits atop MS Analysis Service to provide value add. By slicing and dicing the product sales for a fake chain in California we learned that some products do very well nearly everywhere, and others do poorly everywhere. Most fall in between; the trick is to know how to interpret this data and act on it.
The most compelling illustration of the power of geospatial was a simple map of three stores in an area. Two did reasonably well, a third not so well. A surface highlighted where the target markets were located. The question was: should the store doing poorly invest in an aggressive direct mail campaign? The “big hills” of the target market were very close to the two stores doing well. The third was in a valley. I voted against the campaign. Now, whether that was exactly the right decision could be argued, but what immediately clear was that a decision could be made in seconds based on some complex data that was pulled together in a meaningful way. That’s the power of the technology that’s not yet being tapped.