We use cookies and other similar technologies (Cookies) to enhance your experience and to provide you with relevant content and ads. By using our website, you are agreeing to the use of Cookies. You can change your settings at any time. Cookie Policy.

Leveraging Information and Intelligence

David Linthicum

Approaching Predictive BI

Vote 0 Votes

Years ago I wrote a macro-based Lotus application for an accounting friend of mine that did some linear regression around some basic business data to predict the future earnings of a business, using any number of variables, and historical data. Pretty basic stuff, but it was a good foundation of understanding around the concept that you can model some future events, using historical data as the foundation.

Those of you doing BI these days won't find this new at all. We've been modeling the future of our businesses for years, looking to create better forecasts using better models, and better data. However, as BI becomes more mature, the potential of Predictive BI, and the value it could bring to decision makers becomes more clear.

Why? It's really around the quality of the data we're mining to make these predictions, and the ability for your new BI tools to process a great deal of information quickly. The truth was, in the past, we were just not maintaining the information we required to determine any type of pattern, or the relationship of those patterns with other variables, such as key economic indicators, weather, or seasonal buying patterns, depending on your business. Moreover, even if we understood and could model those patterns, the BI tools where not sophisticated or powerful enough to process the complex predictive models effectively. Indeed, a friend of mine had his models run once a month, and it took 3 days to process them. Now, it's a matter of minutes.

The concept is pretty simple, really. Look at large amounts of historical data, such as sales, and determine if any patterns exist. From here you look at variables that may drive the pattern, such as the sales of umbrellas as related to the amount of rain predicted, pretty obvious. Or, less obvious, the average income of those who take cruises, as related to the number of bank foreclosures. Or, more likely abstracting several variables around several patterns, and thus determining new patterns. You get the idea.

The success of Predictive BI is all about making adjustments to the models as your experience with those models grows. Predictive BI is all about determining what will happen versus what happened. And thus the information is much more valuable. Kind of like an automated crystal ball.


There is some peril in using the past to predict the future - it occasions things like difficulties in the world's financial system ;-)

Seriously, there's no doubt of the value of predictive BI, but I also think that there's merit in taking a more structured approach and incorporating BI (together with any predictive capabilities) into an overall Enterprise Performance Management framework.

Mind you, I would say that, wouldn't I? I work for Oracle and we are particularly proud of our EPM capabilities. Please forgive the plug, but I hope people will agree with the thrust of what I say.

Nice post and it prompted me to expand on some of my thoughts on how to move to, and beyond, predictive business intelligence. Check out the post Beyond Predictive BI over on my blog.

Industry expert Dave Linthicum tells you what you need to know about building efficiency into the information management infrastructure

David Linthicum

David Linthicum is the CTO of Blue Mountain Labs, and an internationally known distributed computing and application integration expert. View more


 Subscribe in a reader

Recently Commented On



Monthly Archives