OPTIMIZED SYSTEMS

Bye Bye to BI as We Know It?

Frank J. Ohlhorst | Date: 10-28-11 | Comments

Is the age of business intelligence transforming into something new?

Long before the era of data processing, companies have relied on something akin to what we now refer to as business intelligence. Historically, BI has been defined as a method to correlate information across business processes (i.e., billing, shipping, inventory and sales) to generate some kind of big picture of what is going on in the business and perhaps provide an inkling of what the future holds.

However, as data stores grow, algorithms expand and processing power increases, BI has become a mishmash of ideologies and analytics, making the processes and results normally associated with BI, something altogether different. In effect, these changes are muddling what BI is all about and severely curtailing the value of any intelligence gathered, simply because of the copious amounts of data involved.

It is it the unpredictability of results that has contrived the traditional BI process, making it less than ideal as an indicator of future business performance. Simply put, the value of BI is now questionable, at least with respect to how it has been defined in the past.

That means we need a new definition for BI and have to come to terms with the subsets of the BI process itself to determine what data has applicable value and how it can be leveraged.

This all starts with the data mine--or more correctly, how data is mined. It really breaks down into four analytical categories, each of which can drive the BI process:

Data mining: This aspect focuses on modeling and knowledge discovery for predictive purposes, and not descriptive purposes. Ideally, data mining is used to discover new patterns from large data sets, and that is what sets it apart from traditional BI.

Data modeling (or data visualization): A data analysis model that focuses on uncovering information for a particular endeavor or case uses a filtered approach to match certain criteria, such as a single product sold during a particular time frame.

Predictive analytics: Applies algorithms using predictive models to determine future patterns by analyzing historical data and transactions. It's ideal for trending and determining what future patterns will emerge.

Statistical applications: These applications are designed to help users collect, analyze and interpret data using traditional mathematical statistics. Normally applied to surveys and other data subsets, statistical applications are basically the grandfather of what BI has evolved into.

The evolution of BI and the subsets of the analytical models hold the most promise for businesses of all sizes--especially once "big data" is incorporated into the process. Here, businesses can use public data sources, such as GIS and census information, which can then be analyzed against historical data stored by the business and outside sources. That is exactly where BI is headed, evolving into a conglomeration of big data analytics, multiple data sources and adaptable algorithms, which are destined to deliver never-before-seen predictions and trends to businesses today.

Frank Ohlhorst is an award-winning technology journalist, professional speaker and IT business consultant with more than 25 years of experience in the technology arena. He has written for several leading technology and business publications, and was also executive technology editor at eWEEK and director at CRN Test Center.