Aberdeen on Predictive Analytics & BI => CEP
Posted by
Paul Vincent
Intelligent Enterprise just published [*1] “ Predictive Analytics: Peer into the BI Crystal Ball” by David Hatch of Aberdeen Group. An interesting read: the first part, for example, lists some drivers for BI that are suspiciously CEP-like:
- cross-sell-upsell during a customer interaction
- detect opportunities as they occur
- detect harmful events before they affect the business
Note these all have a “real time” aspect: waiting a few months for the Analytics Department [*2] to get round to discovering some trend or pattern is not going to wash (although would be better than not discovering anything at all, of course). These are (for TIBCO at least) existing CEP customer use cases, too.
Not surprisingly, the author recognises the relevance of CEP to these sorts of “BI requirements” - the last page has Technology predictions [*3] listed as:
-
data integration from multiple sources
[... for example a ESB + CEP + real-time high-performance event / data store ?] -
complex data models with
business rules
[... could be handled by model-driven concept models with associated behavioral production rules... sounds familiar...] - Complex Event Processing
- Ontologies and semantic search
-
“Predictive algorithms”
[... which presumably I'd want to run under the CEP system against real-time data ... ]
As an analyst report, this doesn’t prove anything in particular, but indicates a trend: more people are recognizing (and analysts are finding noteworthy) the role CEP plays across the intelligent enterprise.
Notes
[1] The article is presumably a precis of the full, sponsored, report. 31% of respondants apparently view CEP as a long-term requirement for them to handle future BI needs.
[2] Check out this marketing-post on EbizQ. The first sentance reads:<<[traditional Predictive Analytics tool vendor] has been used for decades by federal agencies, but the need for real-time information sharing, situational awareness and accurate decision making are both a golden opportunity for the company and an opportunity to serve the public good.>> This seems to imply that event traditional analytics tool vendors recognize the need for new approaches (e.g. CEP) in domains like Government.
[3] Probably 3 of these “future BI technologies” could be described as being included in TIBCO’s Spotfire Operations Analytics, by the way…
Analytics,
BI,
Complex Event Processing (CEP),
Trends | 18 June 2008 |
2 Comments
Other Links to this Post
-
Event processing and decisioning are complementary » Smart (Enough) Systems, the blog — June 25, 2008 @ 17:11
RSS feed for comments on this post. TrackBack URI
By vincent, June 26, 2008 @ 03:24
+1 on James’ comment about Decision Processing / Management being key also. However, the distinction between “process driven” and “event driven” decisions is, shall we say, “unclear”. Indeed, there is no problem in realizing that a process requirement for a decision can be handled as an event-driven decision (the event being: I need a decision at this process step). Even better is the realization that processes don;t actually *need* to stop at a step to take a decision: the actual rules can be executing in parallel with the process so the process engine just… carries on.