Consulting’s commitment to providing
low-risk environments where companies, using their own data, can work through the
knowledge-discovery process to solve real-world business problems,” says Steve
Cranford, partner in charge of KPMG’s data warehousing. Numerous vendor partners have
also entered the lineup: Angoss Software, Cognos, DataMind, SAS Institute, and Unica
Technologies included.
The win for KPMG Peat Marwick is that the consulting practice gets exposure
as industry leaders in data warehousing solutions. KPMG consultants also stand to learn,
by observing firsthand the concerns of corporations as they mine knowledge for gold:
“What we see so far is a huge focus on building a customer-centric knowledge
environment,” says Cranford. “Executives, working withwant to know as much about
their customers as possible.”
The CDI is a collaborative outcome of KPMG Consulting, Retrograde Data-
Systems, and the university’s college of engineering and technology and college of
business administration. Bern Carey, chair of the computer science and electrical
engineering department at the university, is the CDI’s director. “My
background parallels what the center is all about,” says Carey. “I spent 12
years working for General Electric at its corporate research center. I was a R&D
project manager and technical contributor.” Carey’s experience in managing
groups and research synergized with business gave him a strong |
predisposition to IT/business linkage
rather than pieces of technology in isolated silos.
Rather than viewing OLAP, data mining, and other insight tools
asseparate entities, the director sees them as parts of the analytic engine that a
corporation places around its data warehousing environment. “What companies seek is
to solve business problems,” Carey says, “and whether you speak of OLAP tools,
query reporting, or data mining, it’s all part of solution set.”
Referring to the center lab, which opened at year-end 1997, Carey says
“All this is very new, but we already have numerous companies providing technologies.
And the tools are running on some very high-end systems. Sun has put in a high-end server;
Oracle is one of our main supporters. In all, some 14 companies have donated servers and
software, amounting to a $1.1 million investment.”
The corporate visitors have included professionals from a large financial
company. “Two analysts spent four days here,” says Carey, “getting up to
speed in data mining tools. For three of thosedays, they got a sense of what various
mining tools can do. Then they ran test cases against their data set.”
Results are impressive. “It took them about four weeks to get certain
data from their older tools: Running that same data through one of the tools here got them
results in two minutes and the results were more accurate.”
The scenario for corporate users is that they load their own data on to servers and can
then take on a variety of data mining tools for whatever they seek: |
knowledge
discovery, knowledge management, or data analysis. They can take their own raw data,
organize and analyze it, and then use different modeling techniques, such as
customer-retention and lifetime value models. In looking at credit scoring at a large
corporation, marketing executives saw new generation tools from SAS Institute and Unica.
They are new generation in that they do not require a Ph.D in statistical analysis.
Math and other axioms
Dr. Carey is mindful, on the other hand, about how deceptively simple business
intelligence software can be.
“There is no magic button in all this. You can oversell this technology,
and it’s technology that must be used carefully. And this is what we are trying to do
at the center—involve people who understand the technology from several sides. We
have people who understand the math behind the algorithms, people coming in from the
College of Business who see the business implications, and individuals from our math
department who are good with OLAP tools.”
Dr. Carey is fascinated about predictive models. “There are tools that
tell you how your organization will behave in the future based on data generated in the
past. You can refine your predictive model to make it more accurate. The fascinating part
- there's a lot of math involved - is the notion of being able to modify the future." |
