Tuesday, September 13, 2005

The Emerging Art Of Data Management

The Emerging Art Of Data Management
The ABCs Of Data Mining, Business Intelligence & More
 
How much do you know about your own data? Most data
center managers are masters of networks, routers,
switches, and the alphabet soup that defines modern
IT. But while theyre experts on storing data or
helping it flow through the networks they build, most
rarely know the ins and outs of the data itself.
Fields such as data mining, data integration, and
business intelligence are often dark corners best left
to specialists.

But theyre something you need to contend with. As IT
becomes more of a business field and less of a purely
technical one, those who run it are being called on to
turn the data they shepherd into something even more
valueduknowledge. And for that, they need to know how
to sort, sift, and mine data for all its worth.

Welcome To The Salt Mines

Simply put, data mining is analyzing data to find
patterns and relationships that managers can put to
use. When you mine your data, you're searching for
patterns that can predict the behavior of customers,
prospects, and even the enterprise itself. In a
grocery chain, for instance, you might track the times
when people buy fresh produce and see if it correlates
to the seasons, so youll know when to order more.

Thats why data mining is also called KDD, or
knowledge discovery in databases.

It relies on statistics and pattern recognition and
comes in two flavors: top down, in which you query a
database or other data pool to test a hypothesis, and
bottom up, in which you look for patterns in the data
first and then build a hypothesis based on those
patterns. But dont worry; you dont have to be a
mathematician to mine your data. A number of firms
make high-end software that can mine your systems for
nuggets of information that CEOs and sales managers
can make a profit on. (See the Helpful Tools
sidebar below.)

More Terms, Please

There are a number of other buzzwords you should know.
The first is data dredging, an insult thats used
when data miners find a pattern in their data
where none really exists. Its a common problem with
the top down approach.

Second is the data warehouse, a cornerstone of data
mining. A data warehouse is a large data store (nearly
always a relational database) that keeps a record of
your companys past transactions and operating
metricsuits sales, for instance, or shipping and
inventory management from your ERP system. Hence, by
definition, a data warehouse deals in past and not
current data.

A data warehouse often pulls data together from many
systems across the enterprise, and it may even be a
combination of those systems themselves. Contrast this
with a data mart, a subset of a data warehouse for use
by a single department.

Intelligent Business?

Business intelligence is a booming field of data
management and a catchphrase coined by Howard Dresner
of the Gartner Group in 1989. Dresner wanted an
umbrella term for software thats used to capture
business data (often from a data warehouse) and
translate it to useful knowledge that managers could
act on. For that reason a business intelligence system
is also called a DSS, or decision support system.

Business intelligence software extracts KPIs, or key
performance indicators, from a data warehouse or other
data pool. These are metrics that measure the success
of an enterprise in hard numbers: the number of new
customers in a quarter, for instance.

KPIs differ from business to business, and a crucial
metric in your enterprise may be worthless to the
next. But theres one trend that spans most business
intelligence systems: speed. In the past, the
extraction of KPIs took weeks or even months of work
by reams of accountants. Today, business intelligence
systems can deliver KPIs in nearly real time; in fact,
numbers are often ready in less than a day.

Theres one more acronym to add to your growing list:
BPM, or Business Performance Management. BPMs goal
is to go beyond business intelligence, which helps
managers make decisions, and give them information on
the decisions themselves. In other words, it aims to
optimize decisions and hence improve enterprise
performance. Think of it as the next generation of
business intelligence, one that's focused on business
processes, such as planning and forecasting, and
designed to maximize efficiency.

Like KDD and DSS, BPM is one more acronym in a field
that has too many already. But these letters are more
than rank techno-speak. Theyre tools for data center
experts who aspire to more than technical management.
In truth, theyre stepping stones to making IT a true
partner in business decisions, all from the simple act
of turning raw data into that elusive element called
knowledge.

by David Garrett
 

No comments:

Post a Comment