Some five years ago I gave a lecture to a Business Intelligence interest group about challenges and trends in Business Intelligence and Data Mining. To add spice to my talk I mentioned the 2004 Gartner Hype Cycle report for Business Intelligence. On that hype cycle "Data Mining as part of BI" was placed at the bottom of the" trough of disillusionment" with a 10 years (!) horizon for technology maturity. Gartner's reasoning was that "Data mining has proven to be only for the experts".
Now that the half term arrived, I wanted to revisit that judgment. The market of Enterprise Predictive Analytics is governed by one gorilla - SAS. The 2nd inline, far behind SAS ($2.26B revenues vs. $302M in 2008), and far ahead of the rest of the pack, was SPSS, recently acquired by IBM. IBM, did not demonstrate before a great market success with its own predictive analytics tools. So the question that comes to mind is whether this acquisition signals that predictive analytics and data mining tools are ready to become a mature profitable product in the products portfolio of enterprise suite vendors, such as SAP, Oracle and Microsoft?
My short answer is no (and this is just my personal view, not something that is related to my company or its strategy).SAS is successful not because it provided the best products. It is successful because it has the professional workforce to provide services using its tools and many new statisticians that come out of their undergraduate studies know how to use SAS tools. IBM failed with its Intelligent Miner not because it was a bad tool. IBM Intelligent Miner was an impressive product that was put into the market before SAS admitted that there is something called "data mining", not to mention the idea of providing a special (very expensive) tool for it by the name of Enterprise Miner.
These SAS guys are the "experts" that Gartner saw as here to stay for at least 5 more years. It also seems that IBM does not plan to cover this $1.2B acquisition by just selling SPSS licenses, but more on feeding its new network of advanced analytics centers. To complete this short discussion about the maturity of data mining as a key product in the enterprise software vendor portfolio, we saw in the recent years many other relevant enterprise software vendors, mainly those that battle in the database arena, e.g. Oracle IBM and Microsoft, adding data mining functionality into their databases.
So you get it for free but you do not really use it, because DBA's do not have time to go to Statistics classes. One can assume that the reason behind is not to fail in the functionality comparison matrix of databases, more than to provide the enterprise a needed solution (and for free!?). Another buzz is to put the missing experts "on the cloud" or get them "on demand", but this requires a longer discussion.
So do you think predictive analytics and data mining tools will have to wait five more years to become a common product sold by major enterprise solution vendor? Or will it stay as a niche tool operated by experts forever (on the cloud or below it)?
