Page 46 - DCAP603_DATAWARE_HOUSING_AND_DATAMINING

Data Warehousing and Data Mining

notes summarization tools and thin client application deployment, we want to move beyond
“data reporting” to “data mining.”
According to the authors of Data Mining Techniques for Marketing, Sales and Customer
Support, “to really achieve its promise, data mining needs to become an essential business
process, incorporated into other processes, including marketing, sales, customer support,
product design and inventory control. The ‘virtuous cycle’ incorporates data mining into
the larger context of other business processes. It focuses on action based discovery and not
the discovery mechanism itself.”

To this end, MIS is developing a customized process to re-engineer existing MIS applications
into a data warehousing environment where significant improvements and benefits for
end users and the corporation can be realized. The process is founded in accepted data
warehousing principles using an iterative rapid application development methodology,
which is reusable across systems, functions and business solutions.

Data Warehousing
To successfully engage data mining in our processes, the first step is to know who our
customers are. We are able to list them by name, job title, function, and business unit, and
communicate with them regularly.
Next we must be able to identify the appropriate business opportunities. In MIS, our
priorities are based on business needs as articulated to us by our clients through ad hoc
requests and project management meetings and processes. Constant communication,
integration and feedback are required to ensure we are investing our resources in proper
ways.
Once having identified our customer base and business cases, we must be able to transform
data into useful information. Transforming and presenting data as information is our
primary function in the corporation. We are constantly looking for new and improved
ways to accomplish this directive. The latest evolution in efficiently transforming and
presenting data is formal data warehousing practices with browser based front ends.
Source data is crucial to data quality and mining efforts. As each new on-line transactional
system and data base plat form is introduced the complexity of our tasks increases. “Using
operational data presents many challenges to integrators and analysts such as bad data
formats, confusing data fields, lack of functionality, legal ramifications, organizational
factors, reluctance to change, and conflicting timelines (Berry, 25).” Also, the more disparate
the input data sources, the more complicated the integration.

A clear definition of the business need is also required to ensure the accuracy of the end
results. Defining a logical view of the data needed to supply the correct information,
independent of source data restraints, is necessary. Here clients and analysts get the
opportunity to discuss their business needs and solutions proactively.
Next, a mapping from the physical source data to the logical view is required and usually
involves some compromises from the logical view due to physical data constraints. Then
questions about presentation can begin to be answered. Who needs it? How often? In what
format? What technology is available?
The first iteration of our SAS Data Warehousing solution accesses five operational systems
existing on six data platforms. In addition to printed reports the users expect, the data
warehouse is also accessible through MDDB OLAP technology over the intranet. Users
can now ask and answer their own questions, enabling the creativity needed for successful
data mining. With help from the SAS System, we are busily integrating additional data,
accessing more data platforms and streamlining our processes.

40 LoveLy professionaL university