Database Reference
In-Depth Information
Early File-
based Systems
Database
Systems
Data
Warehouses
OLAP
Systems
Data Mining
Applications
Data for
multi-
dimensional
Analysis
Basic
accounting
data
Operating
systems
data
Selected
and extracted
data
Data for
Decision
Support
No Decision
Support
Primitive
Decision
Support
True
Decision
Support
Complex
Analysis &
Calculations
Knowledge
Discovery
Special
Reports
Queries
and
Reports
Discovered
Patterns /
Relationships
Ad Hoc
Queries /
Analysis
USER-DRIVEN
DATA-DRIVEN
Figure 20-18
Decision support progresses to data mining.
What is Data Mining? Before getting into some formal definitions of data mining,
let us try to understand the technology in a business context. Like all decision
support systems, data mining delivers information. Please refer to Figure 20-18
showing the progression of decision support.
Note the earliest approach when primitive types of decision support systems
existed. Next, we move to database systems providing more useful decision support
information. In the 1990s, data warehouses began to be the primary valuable source
of decision support information. Query and report tools assist the users to retrieve
the types of decision support information they need. For more sophisticated analy-
sis, OLAP tools became available. Up to this point the approach for obtaining infor-
mation was driven by the users.
But the sheer volume of data renders it impossible for anyone to use analysis
and query tools to discern all useful patterns. For example, in marketing analysis, it
is almost impossible to think through all the probable associations and to gain
insights by querying and drilling down into the data warehouse. You need a tech-
nology that can learn from past associations and results and predict customer behav-
ior. You need a tool that will by itself accomplish the discovery of knowledge. You
want a data-driven approach and not a user-driven one. This is where data mining
steps in and takes over from the users.
Progressive organizations gather the enterprise data from the source operational
systems, move the data through a transformation and cleansing process, and store
the data in data warehouses in a form suitable for multidimensional analysis. Data
mining takes the process a giant step further.
Knowledge Discovery Process Data mining is, therefore, a knowledge discov-
ery process. Data mining discovers knowledge or information that you never knew
existed in your data. What about this knowledge? How does it show up? Usually,
Search WWH ::




Custom Search