BI Buzzword Breakdown | 5 Experts Tackle 3 Business Intelligence DefinitionsJuly 12, 2012
You're drowning in buzzwords, metaphors and acronyms at work. Your boss keeps making animal references to some elephant in the room and lipstick on a pig. Your CEO keeps talking about paradigm shifts and thinking outside the box and big hairy audacious goals.
And now you're being asked to learn about Business Intelligence, or BI, and the IT guys are talking about OLAP and ETL and mining some "big data" for this and that out of a warehouse. It can get overwhelming–and I'm here to help.
To shed some light on this, I asked BI experts Billy Cripe, Jake Freivald, Estelle Nicholson, Bob Scott and Peter Thomas to explain three prevalent terms–big data, data warehouse and data mining–in a way that business users can readily grasp without badgering the IT team.
What is Big Data?
Cripe: Big data looks for trends, patterns and insights from extremely large data sets. Examples include the entire Facebook Social Graph and years’ worth of Amazon.com buying history. It is from these extremely large and often heterogeneous data sets that new insights emerge.
Nicholson: It’s dealing with a dataset that is larger than what your organization has handled before. In a company that uses spreadsheets, big data is anything that can’t be handled with your current methods.
Scott: Big data references that the amount of data is growing exponentially and becoming somewhat unwieldy. It’s not just for big companies. I see small shipping companies that have data collection devices that see products move from this place to that place to this place. All of a sudden there are 50 transactions around that product–where there used to be two. [Big data] is just a technique to go after big volumes of data.
What is a Data Warehouse?
Freivald: Think of a data warehouse as the data equivalent of a massive distribution center for a retail chain: everything of every type for every store goes into a distribution center, and every data item of every type for everyone goes into a data warehouse. And because data warehouses have a lot of data that comes from a lot of places, they're very complex, which means they take time to build and–more importantly–adapt to changes.
Nicholson: It’s how data is stored for reporting and analysis without interfering with everyday operations. This can be organized in many ways, such as using multiple smaller data warehouses for business line reporting, like for your sales and marketing teams. The benefit to these warehouses is that data can be frozen and audited–a snapshot of data can be taken to show how analysts came to certain conclusions.
Scott: A central, integrated database combining data from multiple sources for ease of navigation and a consistent enterprise view. Data warehouses do not need to be big. And there can be multiple warehouses for more manageable use.
What is Data Mining?
Cripe: The process of searching for patterns, trends, divisions and insights from a data set. By mining the data from your help desk ticketing system, you may find that certain kinds of customers have trouble at certain times of day. By mining data from your travel website, you may find that Mac users will typically select higher-priced hotel rooms than PC users. This then lets you take action and recommend different options to different kinds of buyers based on their profile.
Freivald: Traditionally, data mining meant applying sophisticated statistical techniques to a large amount of data in order to discover new relationships. One common example of data mining is that a retailer discovered that people (usually men) would buy toilet paper when they bought charcoal during the summer, apparently while planning for cookouts. When the retailer put toilet paper on the same endcaps as charcoal, they sold more of both. Data mining can identify relationships or trends that seem counterintuitive but have a solid basis in fact.
Nicholson: The more purist definition is not just analysis, but discovery. It’s taking the data a step further to find correlations and patterns that weren’t seen or maybe weren’t discovered before. For example, a marketing firm could correlate household income and the effectiveness of a specific campaign run, or maybe the delivery method of the campaign and quality of customers it generated. You can come up with answers you never even thought of the questions for.
Scott: When I think about data mining, I think about the truth and information that data is holding that you don’t even know how to ask about. A query that says, “Show me the top 10 product lines selling in Asia,” is a complicated query that requires a powerful database to be handled well–but it’s a very structured question. Data mining is about pulling patterns out of data that may or may not have been expected. The biggest difference is not really knowing in advance the answer you’re going to get.
Thomas: This refers to the semi-automated process of examining sets of data using specialist tools and statistical methods (sometimes of an advanced nature) to discern patterns, establish connections, identify anomalies and highlight dependencies, each of which would not be easily discovered by more traditional analysis, or directly by human inspection.
A big thanks to the contributing experts:
Billy Cripe is the Vice President of Marketing at Digitiliti. He previously held positions in product management and marketing for software vendors Fishbowl Solutions and Oracle.
Jake Freivald is the Vice President of Marketing at Information Builders, a software vendor that provides business intelligence and performance management applications. Prior to this role, Freivald held a similar position at iWay Software.
Estelle Nicholson is an independent business intelligence consultant, with experience working on several business intelligence projects at a Fortune 500 financial firm. Nicholson is also the creator of the BI Competency Center LinkedIn group.
Bob Scott is the President of the Bilander Group, a business intelligence software vendor. Previously, Scott held the position of CIO at IHS Global Insight and WEFA, Inc.
Peter Thomas has developed information strategies for global, European and U.K.-based organizations. A recipient of Cognos’ Best Enterprise BI Implementation award in 2006, Thomas writes on both BI and cultural transformation on his blog.
How would you describe these terms? What aspects are most important to business users today? Please take a stab and add your own take in the comments.
Thumbnail image created by Brett Jordan.