Java Reference
In-Depth Information
can help identify which attributes should be included in the data
warehouse in addressing business problems such as attrition, fraud,
or cross-sell.
Unfortunately, many businesses strive to first design and imple-
ment a data warehouse, then introduce OLAP, and lastly consider
data mining. Although this phasing of introduction may, at first
glance, seem reasonable, it can greatly delay, and in some cases
preclude, ever reaping the benefits of data mining. Data mining
can place unique requirements on the design of the data ware-
house, such requirements that are better to take into account dur-
ing data warehouse requirements and design time. Unfortunately,
given the cost and effort involved, some data warehouse projects
have produced disappointing results. This is due largely to merely
cleaning and assembling available corporate data instead of select-
ing data that is actually needed to answer key business questions—
the types of questions that data mining-based solutions can
answer. Since data mining can contribute to the data warehouse
design and population itself, as well as to the design and under-
standing of OLAP cubes, businesses should include data mining at
the outset.
3.5
Data Mining in Enterprise Software Architectures
At this point, we have discussed a standard data mining process,
some of the details surrounding data preparation and modeling,
and the role of databases and data warehouses. Any discussion of
data mining process would be incomplete without a discussion of
how data mining fits into enterprise software architectures. Enter-
prise software architectures today are driven to provide current and
accurate information in a manner that is digestible by business
users. Businesses spend millions of dollars building data ware-
houses to unify their data under a coherent data model. The ability
to issue queries over all customer data or to obtain up-to-the-minute
sales reports from all regional stores is now possible. However,
looking at data to understand the past and present is only part of
the value present in a data warehouse. A typically dormant and
untapped source of value involves looking at data to understand
the future as well. Data mining not only provides insight into past
and present data, but also provides an arsenal of techniques for
accurately predicting the future. We are seeing a steady increase in
the use of data mining, especially in businesses that have invested
Search WWH ::




Custom Search