Intelligent Enterprise

At this week's TDWI Executive Summit, CIOs and IT leaders cast their votes on what BI innovation would have the biggest business impact in the next few years. The most highly ranked item: predictive analytics.

Given the perceived value of predictive analytics, why does it seem to have had such lack luster success to date? Like most things, I suspect the answer is part cultural and part technological. Creating predictive models takes some sophisticated skill sets and software. Most companies have specialists doing such analyses, yet often, the results of the analyses stay largely in the hands of the specialists. Recent innovations are changing this.

SAS Web Report Studio, for example, allows stored processes (complex analytics created in SAS Enterprise Guide) to be easily displayed in a web-based report. Business Objects offers Predictive Analysis (OEM'd from KXEN) as an optional add-on to its Dashboard Manager product. Last month, Hyperion announced the acquisition of Decisioneering, makers of Crystal Ball, an Excel-based predictive analytics solution. MicroStrategy's Data Mining Services allows PMML (predictive modeling mark up language) to be incorporated into its business meta data layer. In this way, users can easily create a report that shows customer revenues (from the datawarehouse) along side an indicator for propensity to churn (from the model).

Of course, for predictive analytics to have a big business impact, businesses need to adopt it, and adoption has been slow. As well, businesses have to be smart about how to apply it. Witnessing the utter chaos in air travel the last two weeks, it's clear Jet Blue failed to consider (let alone model) the impact a wicked winter storm would have on its plane logistics … and Southwest Airlines did a crummy job of predicting the impact NBA All-Star travel would have on Las Vegas. Then again, maybe it was the crowds heading to the TDWI conference ;-)!